Wednesday, September 30, 2020

 My Go-to commercial VSTs

Some time ago, I wrote an article about the free VSTs that I like to use. This time I will present my favorite commercial VSTs. No, this is not yet-another-post on Serum or Diva. In fact, I felt no reason to join the others for them, since I find much more powerful some of the following than Serum, and I own several analog hardware synths, that makes the need for synths of the type of Diva obsolete.

How many soft synths can someone use in a medium sized track? I would say unlimited, since one could create any kind of layering, or modular combinations (more about that, later).

This is a countdown, presenting the synths I actually use the most. To complete the following list, I would add that sometimes two free VSTs join my commercial ones in my production: Surge and Odin2. But let us go back to our topic.

Here is my top-ten, in reverse order of popularity:

#10. Steinberg Padshop: Although a synth that "comes with the DAW", padshop is very useful, and quick for, you guessed, pads (it's only that Rapid, after 1.8 update, is faster with granular synthesis, so Padshop has a strong competitor here).

#9. Waldorf Largo: Do I need some good brass, without the need to open the Virus? Largo will give me that, and others. People nowdays dislike non-vector synths (even a free synth, Surge, has a vector face). Well, good for me, because then I could get Largo with 75% discount.

#8. Synthmaster One: A very nice sounding synth, helps to build quickly support lines, and comes with excellent libraries.

#7. Thorn: Beautiful synth, quick synthesis, stunning glitches, it offers me synthesis in areas not covered by others. Really. I use it a lot.

#6. Wiggle: Yes, yes, and yes. The Chinese synth is an amazing instrument, with unlimited possibilities. Whenever I need to create a "different" sound, that I know it will attract attention, I just wiggle.

#5. Melda MPowersynth: Although this theoretically is part of MSoundFactory (more about that later), by its own is a very powerful synth and the most easy to use when doing additive synthesis (I do a lot, since I commonly need to invent new plucks for each new track). 3%-5% of the stems have MPowersynth.

#4. Arturia Pigments: The Arturia masterpiece is a brilliant synth, but quite heavy as compared with others, and as such does not find the recognition it deserves in my mixes. But its terrific arpeggiator is always a bonus to my tracks.

#3. Synthmaster 2.9: This "little" synth keeps growing and maturing. It has stunning sound quality. It is not higher in the rank mainly due to its details and rather not-so-stremlined workflow, that make sometimes editing a sound a bit tedious task.

#2. Parawave Rapid: This anti-Serum is by far a superior synth than Serum, having an incredible number of features, being continuously in development, and providing the most efficient way for sound design. It is #2 only because of the outwordly capabilities of the synth that is on #1. Rapid is nowadays around 30%-40% of instances.

#1. Melda MSoundFactory: do not think that its price is high; actually it is really discounted for its abilities. Ladies and gentlemen, this is the King of the soft synths: a modular synth that does everything and in a very fast way (comparatively to its complexity). How about running 10 instances of MPowersynth inside MSoundFactory? And that's only the beginning. Super efficient and super fast. Nearly 50% of the soft instances are MSoundFactory in my projects.

An extra mention to my principal soft drum machine: This is Melda MDrummer, an incredible drum power station. I have incorporated all my samples to it, and nowadays its streamlined processing quickens significantly my workflow.

Honorary mention goes also to Sektor, Vaporizer 2 and Codex, three wavetable synths that I commonly don't use (I find Sektor having ugly FX, Vaporizer having an unfamiliar interface and Codex very limited), but I chose to have them and to keep them, and from time to time, they find their way into my mixes. Also mention goes to Vacuum Pro and Retrologue, which compete for the same, analog-ish, position, the first for the quick buttons (and NOT for its voice doubler) and the second for its depth.

Monday, April 6, 2020

COVID-19 - how many got infected, but never recorded as such?


We all know some friends who underwent the covid-19, without ever been officially recorded as such. We all know also that extensive testing may result close to having the actual number of the covid-19 cases. Therefore, as the data available are enriched by the tests applied to every country (from https://www.worldometers.info/coronavirus/), I did an estimation of the number of real patients from covid-19.




The countries considered are the ones with the highest number of patients (excluding china, which has not released number of tests).

Here is the status as of today:



The next chart is revealing how Germany has demonstrated an outstanding testing approach considering the rest countries.

At x-axis is the tests/pop, at y-axis is total deaths. We keep this information for later.

The basic assumption is that we will use the number of deaths as a de facto knowledge on covid-19 spread. Some deviations exist, but most deaths from covid-19 are recorded well.


The number of Patients reported from every country will be a derivative.

The next table shows the tests committed per total death. It is a sign of “sensitivity” of a country over the expansion stage of the virus. The highest the number, the more precise the results expected.
The next table is the diagnosis of patients per test (e.g. the inverse of tests committed per one positive case). We will keep the number of Germany for later, because it has demonstrated a high number of tests over the lowest number of deaths (see previous table).

Now it comes the first assumption. The next table shows an intermediate value of assumed tests done at the rate of Germany. Why Germany? From the above results, we will use Germany as a model country for doing much in an early stage of the virus (which has also the lowest coverage, as we will see in a following table). The formula for the following table is Population_Country X (tests_GER/Population_GER).

Note that the above number of assumed tests does not take into account the stage on which the testing takes place. We will adjust the above value by using the following two tables. The first table shows the deaths/population for each country. I call this “tolerance”. It corresponds to a de facto social pressure.
Then, for each country, we will calculate their “relative tolerance”, their ratio as above over the ratio in Germany. It shows the magnification of the consequences of covid-19 in a country, over Germany’s one, when late/limited testing is applied - it is used as a "temporal" adjuster (disclaimer: some countries went early in the game, and no knowledge on extensive testing existed, but still, the ratio reflects the temporal magnitude). In other words, the following value transforms any country into a "German-minded" one, in terms of the virus testing. The table is as follows:
We will then multiply the above values with the intermediate assumed tests, to calculate how many tests would have been committed if the sensitivity was the one of Germany.
Now, you remember the table above, termed “diagnosis of patients per test”? We will use the value of Germany to calculate the number of patients. This assumes that if a country had done tests at the pace of Germany, they would have achieved the same rate of diagnosis. Note that Germany has the lowest value, e.g. the most tests per patient. I repeat that table for convenience here:

If we multiply the “assumed tests per sensitivity” table, with the value of Germany for “diagnoses: patients/test”, we get the following estimate for the actual number of patients per country, i.e. including cases which have never been tested. I include also the patients recorded. Note that, again, the calculation assumes Germany being “ideal”, as the calculation produces a number equal the reported number of patients in Germany, i.e. if we assume that Germany “loses” patients, these values should be higher for all (but unfortunately, I haven’t any idea to calculate how much). Also, another factor that messes up real patient data are the different policies applied. Being just a model estimate, the next table does not aim but capture the effect created by late/limited testing on a country. The first column is my estimation on actual cases (equals to "patients out there" + recorded), the second column is what is recorded as of today.


Notes:
  • Germany is selected as "clean" case for the following reasons:
    1. Extensive testing in absolute values ,only second to US
    2. Extensive testing in relative to population values, only second to Italy
    3. Lowest patients/tests, allows considering a "limit" on diagnosing
    4. Lowest "tolerance" (see above), allows for estimation due to "scientific" interest, rather than as a consequence of social pressure.
  • The above factors present Germany as the most "clean" case. The more "dirty" you consider Germany (e.g. inaccurate in estimating the real number of patients), the worse scores are assumed for the rest countries.
  • I received feedback by an Iranian friend that the reported number of deaths is much lower than the real one. I promised to check on their time series data, and try to spot any faking (by comparing with other countries). To be continued, there...

Conclusion

This is a model on how much could have been revealed to other countries, should they have used the approach of Germany. It is not a blame game, as late knowledge was available for Germany to take measures, but only to give an idea on a potential number of positive covid-19 cases, based on the present number of deaths.

PS. Some people insist on believing that in Germany the low death scores are because "it is the young people who get the disease". Really? Have you thought maybe that the many deaths in Italy in contrast, could be an outcome of many-many more cases who have never been reported? 

Tuesday, October 30, 2018


My Go-to Free VSTi

 

As a music hobbyist who uses several pieces of hardware, insofar the synthesizers are of concern, I sometimes rely on free VSTis when it comes to virtual instruments.

Undoubtely, there exist some really good commercial virtual machines over there, but tbh, some good and expressive free VSTis, can compensate for moderate productions.

To distill the prefered free VSTis, I have run in the past two tests, see f.ex. my previous posts here and here. It would be of surprise if my preferred ones were not among those scoring high in those tests.

No doubt that this choice reflects the easiness and the expressiveness of the preferred synths. In addition however, I will use the VSTis that I miss in terms of synth type from my hardware (e.g. I find it easier to tackle FM operators using the mouse in the computer screen, rather attempting to communicate with a hardware synth's LEDs).

To facilitate the presentation I include here two lists. The first list is the synths I will consider for special tasks, so they are most often used as a companion to my hardware. The second list simply extends the first list, and includes all VSTis for which I have implemented midi mapping using my availability of 50 slides and rotary knobs and 29 buttons, coming from my master keyboards (Samson Graphite 49 and E-MU Xboard 49) and my master controller (Omnitronic FAD-9). With 50 slides/rotaries and 29 buttons available, I am able to map between 90%-100% of the most important synth parameters - if not all parameters - for most of the VSTis chosen.

As you may have figured, I prefer tweaking those synths with my hardware slides, rather than using the mouse and the GUI (not that I don't use it at all, of course). To make this work, I have come with specific pages/controller maps to quickly navigate to a synth section, using short colored labels grouped by functionality. In the beginning such an approach may require a learning curve, but once mastered it can become more convenient than having to navigate f.ex. to each and every synth's small/large/non-resizable GUI having to use every time different locations for every section, etc.

To get an idea on how such a mapping looks like, see the photo below.

 

The arrangement of the various slides/rotary/buttons corresponds (mostly) to the locations of the hardware as you can see at the photo above (e.g. upper two left sections is the X49, upper two right sections is the FAD9 and the middle left and right section is the Graphite. Bottom line is the buttons, left for the FAD9, and right for the Graphite). You may see in the paper charts that I tend to use similar controller locations (as possible), so f.ex. I don't need to search where is the 2nd ADSR or the Resonance.

So, here's the list of the chosen ones (in alphabetical order). In brackets, it's the parameters that I've mapped.

List 1 - 12 synths that will be my choice #1 depending on the task 


I most often use these irrespectively to having them midi mapped (update:June 2019).

Cobalt (wavetable, linear FM, PWM) [48 parameters + 16 buttons]
Dexed (DX7 emul., FM synthesis 6 OP) [no midi mapping]
Dune CM (wavetable, unison) [32 parameters + 12 buttons]
Oatmeal (VA, wavetable) [no midi mapping] 
Ob-xd (VA,  Oberheim emul.) [39 parameters + 19 buttons]
Superwave P8 (VA, vintage strings) [45 parameters + 7 buttons] 
Surge (wavetable, FM synthesis 3x3OP, mod.matrix) [no midi mapping yet]
Synth1 (VA, Clavia Nord Lead II emul.) [48 parameters + 9 buttons]
ToneZ (4 Osc wavetable, unison) [no midi mapping]
Unifyer 3 (mono, wavetable, unison, arp) [no midi mapping]
VK-1 Viking (mono,VA, Moog emul.) [no midi mapping yet]
Xhip (VA and PCM, waveshape, mod. filter) [48 parameters]


List 2 - 20 synths that I have fully midi-mapped

Midi-mapped synths, because I consider them useful (and they support the easy way to map a VSTi, that is midi learn). Some of the following ones appear in the top-10 list above, too.

Virtual Analog-Digital, Wavetables
1. Alpha CM (wavetable, 2x2 Osc, quick editing) [32 parameters]
2. Athmonova (strings, pads, unison) [32 parameters]
3. Cobalt (wavetable, FM, PWM) [48 parameters + 16 buttons]
4. Dune CM (wavetable, unison) [32 parameters + 12 buttons]
5. Superwave P8 (VA, vintage strings) [45 parameters + 7 buttons]
6. Synth1 (VA, Clavia Nord Lead II emul.) [48 parameters + 9 buttons]
7. Swierk (VA, 2 x everything, mod. matrix) [48 parameters + 4 buttons]
8. Tyrell N6 (VA, ring, mod. filter, mod. matrix) [48 parameters + 5 buttons]
9. Xhip (VA and PCM, waveshape, mod. filter) [48 parameters]

FM
10. Brzoza (FM synthesis 2x3OP) [50 parameters + 12 buttons]
11. K200 (FM synthesis 4OP) [50 parameters + 5 buttons]

Additive - Spectral - Waveshape - Phase distortion
12. Lynx (Wavedraw synthesis, phase distortion) [16 parameters]
13. Zebralette (Spectral synthesis, mod. matrix) [50 parameters + 5 buttons] 

Vintage - virtual replicas
14. Gigmate v2 (VA, Concertmate MG-1 emul.) [21 parameters + 5 buttons]
15. Mono/Fury (VA, 4 Osc, Korg Mono/Poly emul.) [32 parameters]
16. Ob-xd (VA,  Oberheim emul.) [39 parameters + 19 buttons]
17. Phutura (VA, Roland Alpha Juno emul.) [29 parameters]
18. Poly-2016 (VA, Roland Juno-106 emul.) [35 parameters + 16 buttons] 
19. Prophanity V1-5FE (VA, SC Prophet-5 emul.) [24 parameters + 16 buttons]
20. U-No-62 (VA, Roland Juno-60 emul.) [20 parameters + 10 buttons]


There are many other possibilities (f.ex. I use physical modelling or formant synthesis as well), when it comes to sound and music synthesis using VSTis. The above template simply suits my regular needs, but it is nevertheless filtered in terms of quality, diversity and capability for those free VSTis, as well as in terms of sound peculiarity (this concerns the replicas). That said, do not underestimate some older free jewels from the world of the VSTis - at least in my case (where I mostly use my hardware), they have pretty often saved the day.

About me


I'm a music hobbyist enjoying writing and instrumenting music in my spare time. My involvement with the music begins since 1985. Among others, I bought my Roland SH-101 in 1986, my Yamaha TG-500 in 1992 and my Clavinova and SY-35 in 1994. I still have (and use) the first three ones, together with my Access Virus TI2, Clavia Nord Lead A1, Novation Peak, Behringer DeepMind 12, Elektron Digitakt, Roland SH-201, Roland JV-1080, Behringer Neutron, Behringer Model D, and my software (VST) instruments.


Saturday, July 28, 2018

Download free GRADIENT and MEMFIS for Research Use


This is the page to download Gradient and MEMFIS. Both are very powerful regression systems, which use grammar-guided genetic programming. Gradient produces ensembles of neural networks and support vector machines, while MEMFIS uses a memetic approach to calculate precise Takagi-Sugeno neuro-fuzzy systems. More information on Gradient can be found here, and for MEMFIS can be found here.

Both systems have been produced as part of the EU-funded project INFER. In addition, I received funding (senior research fellowship) by the Marie Skłodowska-Curie Actions - European Commission.

Gradient and MEMFIS are provided as standalone Windows applications (they should run without problems to any 32-bit Win system, e.g. XP,Vista,7,8 and 10, presumably to 64-bit too, but this is not yet tested).

You may use freely Gradient and MEMFIS for your research, and freely publish your results, using the following citation(s):

  • For Gradient:
Athanasios Tsakonas, Bogdan Gabrys, GRADIENT: Grammar-driven genetic programming framework for building multi-component, hierarchical predictive systems, Expert Systems with Applications, Volume 39, Issue 18, 2012, Pages 13253-13266, ISSN 0957-4174, https://doi.org/10.1016/j.eswa.2012.05.076. 
  • For MEMFIS:
Athanasios Tsakonas, Local and global optimization for Takagi–Sugeno fuzzy system by memetic genetic programming, Expert Systems with Applications, Volume 40, Issue 8, 2013, Pages 3282-3298, ISSN 0957-4174, https://doi.org/10.1016/j.eswa.2012.12.099.

No responsibility is assumed from my side from the result of these systems.

Please contact me first, if you plan any commercial use.

What is included in the download files


Each download file contains a directory with the regression system executable (and any supporting files), and another directory where a 'Profiler' software exists, that allows to update the parameter files (e.g. it's just a one-page software menu with data entries, so you don't have to manually update - and guess - the numbers included inside the parameter files). The systems have been 'tuned' to default/preferred settings. Use any other setting at your own risk. The download file contains also several DOS batches which will execute run/search on some default public domain data/problems. I recommend you run first the batches so you get an idea. Then install the Profiler, and open one "syspar" file, so you get another idea on the tuning. At first, it is recommended that you keep the default settings for your problem and start tweaking only when you feel you may improve the results. The programs create sub-directories and files, so make sure you have admin rights on the chosen directories.

  • Download Gradient from here.
  • Download MEMFIS from here.

Friday, July 20, 2018


25 Labels, 123 free VST instruments and an Ambient Song


Introduction 

It's not much time since my previous review on 32 general-purpose synths, and I am ready to present the second one. This time it's not individual synthesizers that compete, but combis of synthesizers provided by a single label (provider) at a time. In other words, for a test song (i.e. an ambient song) 25 different versions were created using each time 5 free VSTis from a different label/developer/company. Each version is available here to let your ears to be the final judge.

I will discuss my experience as I try to fit the needs of an ambient song into the provided synthesizers by a single label (developer), at a time. This aims to be a test that would measure the efficiency and the effectiveness of the whole process as seen from within the realm of a VSTi developer. Individual synthesizer capabilities will also have the chance to be discussed.

It is often considered that synthesizers that are produced by the same developer often share similar capabilities or similar defects, which has given good or bad reputation to some of them. Let us find out whether this is true, with this opportunity.


Setting everything up

 

The song

The song is an ambient/downtempo song I've written, it's called "Amber" and it was especially composed for this review. For its original version, 9 individual synthesizers from various labels were semi-randomly selected. The exact synth configuration of the original version will not be disclosed now but it is enough to say that if any of the synthesizers used in the original version was appearing into one of the combis (and you may probably recognize some, after hearing them), its performance was not taken into consideration by me, during the evaluation.

"Amber" was intentionally constructed using 8 melodic lines that interchange in groups of two, and 1 texture line, in order to allow for easy evaluation. Each group of two lines will henceforth be called as a "Role". Therefore, in Amber there are 4 melodic Roles and 1 Texture Role to be distributed to 5 different synthesizers or to 4+1 synthesizers (i.e. 4 different synthesizers for the melodic line plus 1 more instance of one of the 4, for the texture - see however later for details on Roles distribution). 

Thus the four Roles are:
  • Role A: consists of two lines:
    • Line 1: Bell piano/Ambient piano/Toy piano. This is a piano with a celesta element, with some delay (optional) and lots of reverb (mandatory), polyphonic.
    • Line 2: Phasing contrabass: A paddy bass line with sweeping elements, may include ramp and/or saw waveforms, but not too sizzly, polyphonic.
  • Role B: consists also of two lines:
    •  Line 3: Chorus pad: as the name implies, a potentially analog chorus pad, preferably female (higher tone), polyphonic.
    • Line 4: Cello string: a potentially realistic cello, or vibrato on saws with semi-open 24dB filter and half-open envelope filter level, can be phased or FM-ed, monophonic.
  • Role C: two lines as well:
    • Line 5: Rough contrabass: a slightly screaming contrabass with vibrato and possibly early reflection reverb or delay, monophonic.
    • Line 6: Evolving pad: a warm pad where the filter is modulated by a S&H LFO or similar, potentially introducing slight resonance, it should at least incorporate some time modulation, polyphonic.
  • Role D: here also two lines:
    • Line 7: Cold flute: an analog sine or square wave flute with slight white noise, medium attack and release, medium filter envelope, plus a bit of portamento, monophonic.
    • Line 8: Toy xylophone: Pseudo-arpeggio of an analog sine or square percussion, e.g. the arpeggio has to be created by single notes, or LFO+ENV trig., or in worse case a delay with lots of feedback, monophonic.
  • Texture role: here only one line:
    • Line 9: The brook: a quite water flow from a small stream is expected, or a white noise filtered by LP and HP, or if only one filter exists, by a BP, or in the worse case, just some (filtered) white noise. It can get even worse or course, with some phase distortion synths, where the noise will be produced by the lowest possible note, adding some freaking modulation to it.
As it can be seen, apart from Role D (more analog, monophonic role, and targetted to pseudo-arpeggios) and the Texture role (noise manipulation, analog too), the rest three roles include mixed requirements. This was chosen intentionally, in order to attempt to push the synthesizers to some limits, e.g. stretching from a string synthesizer a "chorus pad", or getting from a physical modelling instrument a "phasing contrabass" would have shown versatility on the part of the synthesizer. Actually, some of the tested VSTis performed exceptionally well, given their original purpose.


The environment

 

The same configuration as in my previous review is used. The DAW is Cubase Elements 9.5 64bit, running on an Intel Core i5-7200 @2.50GHz having 8GB RAM using a Focusrite Scarlett 6x6. The 32bit synthesizers were bridged to 64bits using JBridge. So the hardware is rather to the high end for 2018, and the performance of the software synths should be expected good.

All instruments in "Amber" pass through Steinbergs Parametric EQ (with preset but custom settings per each Role).

Effects in the mix

 

The following effects section is applied to the mix (in this order): Steinberg Compressor, LKJB Psy Warmth (saturator), Steinberg Maximizer, Steinberg UV22HR.  This is a pretty light configuration, but whenever considered that it is needed, Steinbergs Reverb, Delay and Ping Pong Delay may be added (to single VSTis).

In "Amber", there are also a few additional sound effect waves, either played in real time by one non-competition noise synthesizer (which features low CPU consumption), or as prerecorded sounds. The same sound effect waves are used in all versions.

 

The Synth Rack - what is going to be rated?

 

- As mentioned previously, a combi of 4+1 synthesizers, from a single VSTi label, at a time, will be rated.

- Melodic lines. Namely, 4 different synthesizers from a single developer must be used for the melodic lines of the song. This means that in order to qualify for this competition, a developer has to have produced at least four different free VSTis. These 4 VSTis will support 8 melodic lines in total, hence each of them must support two alternating lines, receiving program change messages.

- Texture. 1 additional synthesizer from the same developer will be used for producing ambient texture (just 1 line), however this can be a second instance of the above four, if (and only if) the developer has released only 4 free VSTis (otherwise, it must be a different VSTi of the same developer).

- In summary: Combis of 5 synths or (4 synths+1 more instance) will be tested.


Note that labels providing less than 4 free synthesizers, or providing known CPU-hogs or otherwise problematic synths, or with mainly lo-fi or toy synthesizers, are not taking part in this review. This does not mean that if some labels are not listed here, then they belong necessarily to one of the categories above.



How are the versions going to be rated?


The following rules were made aiming to provide for a fairer comparison between labels. For example, some labels have numerous synths which would give them advantage over others, should there were any efficiency issues. But this review aims to draw an impression by randomly choosing one subset from a provider, and not their whole contribution to the free VSTi market. Therefore it limits the participation of any label to a chosen subset of VSTis. Random choosing here means that I don't know a-priori what is expected from the chosen synth, and not that I blindly pick up some. In some cases, and if I have the option, I intentionally include some synths with known characteristics from the provider, e.g. a synth with arpeggios for Role D, some analog for noise, etc. But once a selection is made, I expect that the label will not oblige me to make changes (because f.ex. a synth of them doesn't output sound).

The two qualities

Each song version will be rated in terms of (a) effectiveness and (b) efficiency. Effectiveness is the ability in reproducing the song. Efficiency is the easiness and the practicality (e.g. low CPU demand) of the synth combination. More specifically:

Effectiveness

- Get the sound. Each sound (i.e. each melodic line and the texture) is independently evaluated in terms of equivalence to the original (e.g. timbre and overall) and the result from all Roles is averaged.
- What is this sound? Structural deviations in the sounds (technical deviations, e.g. missing special characteristics of the required sounds [like not being able to produce an arp/delay effect]) will have penalties in effectiveness.
- What is this song? Deviations from the sounds which change even the character of the song as a whole (aesthetic deviation), will have further penalty in effectiveness (e.g. a result were the song sounds very '70's).


Efficiency

- (No) time limit. A maximum of 3 hours is set as the time limit to configure each synthesizer group. Faster setups do not get bonus, while a time longer than 3 hours may also be considered, but it adds an efficiency penalty. In most cases however, no more than a 1.5 hours were needed for the complete process, but the time limit was allowed to be stretched since I couldn't have the same familiarity with any of these VSTis, while it is also true that for some for synth types more manipulation/experimentation is required to get to the result.

- No CPU hogs. Any high CPU demand and/or producing 100% CPU spikes during play, will have penalty in efficiency. This is because we test free VSTis, commonly used by people who cannot afford fast computers - and which are supposed to be mainly lightweight ones, if it is expected to be used for music production. For the requirements of this song, and the specifications of my computer, a CPU consumption up to 50% is considered good. Lower CPU values are welcome and they will be reported, but they will not get bonus (since there is no actual technical effect for this song).

- No extra VSTis, please. Related to the above reason, any obligatory use of an additional instance of a synthesizer during setup (f.ex. because program change messages are not supported by the VSTi), will have penalty in efficiency. If this results also in significantly higher CPU, further penalty is applied. [In this ambient song, failing to support program change messages requires 2 synth instances for a single role, since one role has two melodic lines that alternate. What would be the CPU demand if one role had four different melodic lines, for a synth that doesn't support program change messages?].

- No different synths per role parts. As an additional constraint, if it is required an extra synthesizer instance for a line (e.g. because program change messages are not supported) this will be a second instance of the synthesizer that has been already used for that role, and not any other synthesizer (or other instance). Exception is if there is a multi-timbral synthesizer that could cover the missing instance using a second channel, for example, saving the need for more instances. 

- No added FX plugins. Any use of another, additional plugin (e.g. reverb, delay, filter) which will be considered mandatory from me (i.e. to reach an appropriate level in quality), will bring some penalty in efficiency (but it will obviously at the same time improve the effectiveness rating).

- No crashes!. Any realized VSTi crash, or obligatory replacement of a synthesizer because of incompatibility, any other obligatory change in the initial synth configuration that I will encounter,  or any other malfunctioning, which affects workflow and/or quality will have penalty in efficiency.

What is not going to be rated?

 

- The GUI is not going to be rated (e.g. fancy, realistic, simple, etc.), unless it results into a severely hindered workflow (e.g. knobs totally messed up), which I don't expect to be the case.

- Other technical features, like size in the disk, installation, the existence of keyboard, etc. are not important for this review.

- Midi learn, midi controls responsiveness, keyboard tracking, etc., although important to validate the value of a synth, are not part of this review (I just didn't have the time for such a thorough testing now - but perhaps in the future).


To recap: why these rules and why this song?

 

The main reasons for selecting the above rules are the following:

- Good tester: An ambient song has often very diverse demands in sound design, and sometimes it's even only the sound, and not the melody that makes a song very special. So, I consider that an ambient song makes a good tester for a group of synthesizers. On the other hand, it's not so difficult to create the sounds or equivalents, and it is neither biased too, like it would be f.ex. if realistic electric guitars were needed. Perhaps the classic virtual analog or vintage analog synths will have less success for this ambient song in general, but nothing prevents a decent pad from a vintage synth, since many of the required elements can be found there.

- Effectiveness: The four roles for the melodic lines of the synths have been selected so as to push the synthesizers to some limits and not just picking the best of every synth. These limitations obviously require a good planning on which synthesizer will take which role.

- Efficiency: To support song-making and reduce the CPU demands, a synth should support program changes so the same instance can be reused inside the song. Furthermore, program changes must be fast and smooth (e.g. without FX residuals mirroring to the next patch, audible gaps, missing notes, etc.) and the generic CPU consumption should be low and by all means to never reach 100%.

 

Who is competing who?

 

The labels that will produce the versions of "Amber" are the following ones, in alphabetical order.

  • Acrobatics
  • Algomusic
  • Angular Momentum
  • B.Serrano
  • Big Tick
  • Contralogic Productions
  • DSK
  • EFM
  • Electrostudio
  • EVM Synths
  • Full Bucket Music
  • GTG
  • Krakli
  • Kriminal 
  • KX77FREE
  • L-Day
  • Novaflash
  • Novakill
  • Odosynths
  • Psychic Modulation
  • Saltline/Drzewo
  • Syncersoft
  • Togu Audio Line
  • U-he
  • Xenobioz
Each of the labels will be presented in the paragraphs that follow, also in alphabetical order. In every paragraph, I discuss my experience while fitting the song to every synth combi, and I present the derived rating.


The Preparation



Extended search has been commited to locate and install the various VSTis tested. The selection of the labels was not easy as well, since there are many providers out there who have 4 free VSTis. The criteria was both the popularity and the diversity. The final 25 labels that made it to this list, have left about 10 other candidates out.


The Execution


Although the results appear here alphabetically, the song versions were created in a completely random order. It required around 1 1/5 month overall to complete the review, since I could work on this project only in my spare time.


The results




Note that in regard to the rating that follows, a serious attempt was made to assess fairly each competitor. This included repetition over repetition of the assessment, and pairwise comparisons for closely rated labels, updating half a point here and there, checking overall averages and standard deviations, to feel sure that the rating was fair. It may not be possible to avoid a mistake, but by releasing the results of this assessment I feel confident that they represent an objective judgement from my side in respect to the posed rules and the realised performances.

But first, a special hint: 

 

For each label, you may also listen in the following paragraphs their produced song version so you may make your own opinion. I strongly recommend however that you don't listen subsequently two versions but let at least several minutes in between. This because you might prefer to have your ears a bit rested after listening to one version, in order to listen to the second without any bias residing from the previous hearing. I promise you, your listening experience and judgment will be better this way. Just give the chance to each version for an equal listening treatment - each label provides each own sound signature.
But first you might want to listen to the original "Amber" composition so you have a measure of the aimed goal (which, original version, perhaps will not remain the "best" version afterwards).
You may listen to the original version of "Amber" here.


Acrobatics


This label has released a minisynths series, but it has been mostly known for their unique Alieno ambient/experimental synth. While setting up the song, the first thing that I realized was that Acrobatics VSTis do not support program any change messages, requiring double instances for each Role. The label claims that their synths include optimized code. Will they manage to avoid spikes with 9 instances ? Let's see. For Role A, the phase modulation Fret had no problem to provide good sound interpretations. Grit, a pulse saw analog undertook with sucess Role B. Role C was given to the analog Peak which wasn't shown much powerful, but it managed to provide some adequate output. Role D was assigned to Mate, which is not part of the minisynth series, but an (abstract) vintage synth emulation. Mate was shown pretty much powerful in this Role. Finally, the Texture was taken by Alieno which proved it was peanuts for it. The song version was very good, verifying that Acrobatics VSTis have a good sound. I have added some reverb to Peak and Fret (the latter didn't need it really, but I felt I should add it). Despite the 9 instances loaded, the CPU remained between 40%-50%, having only a single spike of 100% CPU during live play (honestly I was expecting a worse CPU performance since these synths have been created by SynthMaker). For the effectiveness it's a 3.0 counted, and for the efficiency, due to the many instances required (e.g. no program changes support at all), the single 100% CPU spike, and the added reverb, it's a 1.5.
You may listen to Acrobatics' version here.


VSTi
Year
Bits
Rating
Role A
Fret (A) / Fret (B)
2012
32
VeryGood
Role B
Grit (A) / Grit (B)
2012
32
Good
Role C
Peak (A) / Peak (B)
2012
32
Fair
Role D
Mate (A) / Mate (B)
2014
32
Good
Texture
Alieno
2014
32
VeryGood

 

Algomusic

The legacy VSTi provider has released several significant free and commercial synths in the past. Unfortunately, the production has ended for some reason, and software code has been lost, in order for someone to continue development. For the needs of Role A, the phase distortion Czynthia has been selected, a VSTi based on Casio CZ, which actually sounds really good. Role B was taken by StringSynth, an emulator or Arp Omni-2 vintage strings synthesizer. Role D was given to M42 Nebula V.2, a pads synthesizer, which was shown more than capable of providing the basic - for Nebula - sounds. Role C was assigned to the larger brother of Nebula, the M51 Galaxy, which however demonstrated difficulties with program changes being so slow that they were resulting into audible gaps, enforcing me to move the program change messages into more safe timings (which was not always an option, though). For the Texture, the options were running low, and I had to use the phase distortion Phadiz, which doesn't have a noise source and it had to be really anorthodoxically squeezed to provide the texture you will hear. Although the CPU was mostly between 30%-50%, a 100% CPU spike could not be avoided at the moment the synths were changing programs which often created also failures in the program changes themselves. Furthermore, some failures in program changes were partially healed when I used exclusive channels for each synth (I don't know why the routing was causing interference before). In addition, for some reason also unknown to me, after recording the song, Czynthia fails to change from the bell piano. Overall, Algomusic's aged VSTis are generally rich and capable in sound, but only M42 Nebula V.2 and Phadiz demonstrated some reliability under production terms, and for this, I also cannot be sure whether it was accidental or not. But the song version was smooth and sweet, apart from the monotonous piano of Czynthia denying to change patch in the beginning (it made it though, later). For the effectiveness, concidering that I want to hear what I have programmed to, the rating amounts sadly to only a 2.0 and for the efficiency it is a 3.5 (a CPU spike, slow changes, issues).
You may listen to Algomusic's version here.


VSTi
Year
Bits
Rating
Role A
Czynthia
2009
32
Good
Role B
StringSynth
2004
32
VeryGood
Role C
M51 Galaxy
2006
32
Fair
Role D
M42 Nebula V.2
2006
32
Good
Texture
Phadiz
2003
32
Fair

 

Angular Momentum

The label has four free synthesizers available, two additive/wavedraw ones, an analog (which focus mainly on bass sound) and a wavetable. The analog, Analog Warfare, for its S&H and sequencing functionalities was assigned Role D. Role C was given to Retronix, an impressive wavetable, while Role B was given to WavedrawFree, to let the Role A for the Freehand. Since the developer has not other synthesizers available, the Texture role was given to a second instance Retronix. From what I figured, one synth (Analog Warfare) has been developed using SynthEdit, and the rest with SynthMaker. So I got worried, when I realized that Freehand, WavedrawFree and Analog Warfare were really ignoring program change messages, raising into 8 the needed instances (Retronix responded fine so only one instance per line was used). However, it seems that these VSTis have very optimized code, or that they use custom code, because the CPU never went above 35%-40% compensating a bit for the overhead. They produced easily and quickly a nice version, and the developer would have been competing for the top here, should they were supporting patch switching. Since one of them, Retronix, does support this, it is also recommended here for further exploration. The resulted rating is 3.0 for the effectiveness, and 2.5 for the efficiency, which already includes a bonus - given that no patches were changing - because of their low CPU consumption.
You may listen to Angular Momentum's version here.


VSTi
Year
Bits
Rating
Role A
Freehand (A) / Freehand (B)
2007
32
Good
Role B
WavedrawFree (A) / WavedrawFree (B)
2010
32
Good
Role C
Retronix
2010
32
VeryGood
Role D
An.Warfare (A) / An. Warfare (B)
2007
32
Good
Texture
Retronix
2010
32
VeryGood

 

B. Serrano

The synthesizers of B.Serrano are characterized commonly by advanced routings which often result into interesting and unique sounds. There is a substantial number of free VSTis from the French developer to choose. For Role A therefore, I selected the flagship (to my understanding) Anubis-2, which was proven a very capable synth featuring very nice (e.g. modern) sound. Role C was assigned to Orpheus-II a string synthesizer, for which it had no problem to accomplish. Horus, another string synthesizer, was given Role B, when the first selection, the FM Adonis crashed when I loaded it in my DAW. Radium took effectively Role D, and the Texture was handled straightforwardly and very realistically by the analog Artemis. Unfortunately, this configuration resulted into regular 100% CPU spikes, while there was also audible latency on program changes, in some cases also recorded in the final recording. In addition, during setup, Radium and Horus were occasionally encountering some muting situations, and were stopping to produce any sound, requiring from me to reload the plugins. Overall, B.Serrano has provided us with synths featuring rich sound and many options, but the functionality and reliability of some of the synthesizers does not stand at an equal level, plus the high CPU consumption prevents these synths from being used easily into larger synth group setups. The rating for effectiveness counts a 3.5 (e.g. missing notes due to patch changing) and for efficiency is a 2.5 (many CPU spikes, audible latencies, crash of Adonis, etc.).
You may listen to B.Serrano's version here.



VSTi
Year
Bits
Rating
Role A
Anubis-2
2015
32
Good
Role B
Horus
2014
32
Fair
Role C
Orpheus-II
2016
32
Good
Role D
Radium
2011
32
Fair
Texture
Artemis
2007
32
VeryGood

 

Big Tick

This legacy label has released distinct VSTis as free or commercial ones, focusing in the quality of the sound and the efficiency as well. For the Role A, their tiny EP Station was chosen (tiny only in file size, that is, but great otherwise), which it was shown also capable to handle the non-piano requirements of this role as well. Role B was given to Rhino CM, which is not officially a free VSTi, however it comes for free along with another 1.5Gb or numerous VSTis from CM Magazine for a total cost of one magazine issue, which is around $5. The Cheese Machine undertook Role C, for which it was more than capable. Role D was given to the FM synth Rainbow 2, which however was not responding to any program change messages, requiring double instance for this role. Running out of options again, and the Texture had to be given to the heroic formant synthesizer Angelina, which did its best, satisfying at least me with the result. The only thing needed extra to Angelina was a TAL Filter, to shape a bit more the result. A reverb was added to EP Piano, too. Rhino CM did not perform correct program changes since the parameters in its modulation matrix would not load/save (some issue with the CM version?).  Angelina crashed the first time I loaded it in my DAW, but mysteriously, did not have any problem afterwards (I always test twice a crashing plugin having read about chaos theory).  Overall, for the sweet version it provided, Big Tick gets a 4.0 in effectiveness, and a 2.5 in efficiency.
You may listen to Big Tick's version here.


VSTi
Year
Bits
Rating
Role A
EP Station
2003
32
Good
Role B
Rhino CM
2010
32
Good
Role C
Cheese Machine
2003
32
VeryGood
Role D
Rainbow 2 (A) / Rainbow 2 (B)
2002
32
Fair
Texture
Angelina
2010
32
Good

 

Contralogic Productions

This label has released a group of different-tasked VSTis, such as vintage emulators, supersaw analogs, etc., in a span of 10 years, where each of them has some features that make them special. Generally, their synths sound very good. The newer synths demonstrate a further progress the developer has undergone over the years. For the needs of the song, the flagship Feldspar, a time modulation VSTi, was assigned Role C, since a nice rythmic pad was in need. The choir-stringy Role B was given to the saw analog Pterosaur. Both synths demonstrated smooth program changes and flawless functionality, providing quality sound. Leto, an emulator of the vintage synthesizer Synthetone SX-1000, provided the Texture, an easy task for such a VA. Now, Role A was given to the FM-synth Cadencia. Cadencia has a nice and rich sound. However, this old FM does not provide support for program changes, requiring two instances for my Role. Moreover, it leaves ghost notes, and noticeably in the ending of the recording (which will cost to this label rating in the efficiency, and not in the effectiveness), where a note remained there playing, "hanging" till the end of the recording. Additionally, the bell piano of Cadencia required a ping pong delay for the ambient flavor since its built-in effect, "Effect", altered the pitch of the note when its depth or its feedback tuning were large. Foxglove, a morphing synthesizer, as a monophonic, took role D, for which it had no problem, except that it was also not providing program changes, requiring two instances. The result was melodic, and the CPU consumption, despite the 7 instances, remained fair, between 40%-50%. Overall, the newer synthesizers (e.g. 2007 henceforth) from Contralogic Productions are reliable and they provide good sound, while the two older ones may have some functionality issues, although Foxglove might still be a consideration for some tasks. For the effectiveness, it's counted a 3.5, and for the efficiency it's a 2.5. Note, that the same label has provided us also with a straightforward and practical groove machine, RPS-1, worthy maybe to have a look.
You may listen to Contralogic's version here.



VSTi
Year
Bits
Rating
Role A
Cadencia (A) / Cadencia (B)
2003
32
Poor
Role B
Pterosaur
2012
32
VeryGood
Role C
Feldspar
2013
32
VeryGood
Role D
Foxglove (A) / Foxglove (B)
2003
32
Fair
Texture
Leto
2007
32
Good

 

DSK

This is a contemporary free (mostly) VSTi label, that has produced almost all kinds of synthesizers/romplers, often coming with different variants/approaches for a specific task. This setup for the song aimed to use solely the wavetable synths of DSK. For this reason, Synthopia 2 undertook Role A, and Choirz handled Role B. Role D was given to ZitH, while Ethereal PadZ 2 was selected for Role C. In the Texture, it was the specialized Darkness Theory. Although programming with the wavetables is supposed to be a fun process, in the case of DSK it got me really frustrated. For some still unknown reason, during play some instruments would go completely out of tune, turning the song into a cacophony. The problem was spotted in Ethereal PadZ 2, a 3-layer wavetable pads synth, where one layer had its own life, setting its pitch to zero, in a music part where only notes existed. I removed the synth (completely, that is, from my PC), and I substituted it with Ethereal PadZ, an earlier version, which has only two layers (and therefore lacks a 3rd layer to detune). The scheme worked for a while, but then I realised that the same detuning side effect (defect) was appearing to Synthopia 2, just not so often (an easter egg???). Nevertheless, in the final recording this detuning was not recorded, and I decided to end any further synth configurations. Darkness Theory didn't allow for program changes, but they weren't needed anyway for its role. I only added some extra reverb to it. Another defect I encountered was that occasionally Ethereal PadZ and ChoirZ would lose the preset configuration, requiring from me to reload the patch. Overall, the DSK configuration turned into a frustrating experience with bugs popping up here and there, diminishing my reliability to these VSTis. The song version carries however good elements, and theoretically these wavetables may be a good companion, but the cost in setting, re-setting, withdrawing potentially, and re-configuring them is so high that prevents me from considering of using them in any serious setup. On the other hand, the CPU consumption remained at 50% at maximum, which I consider satisfying for a group of 5 wavetables not featuring any code optimization. For the effectiveness, the rating amounts to a 3.0 (e.g. there are audible spikes on program changes) and for the efficiency is a 3.0 (e.g. at least one of the synths, ZitH, demonstrated reliability). Note, that a slight detuning that you will spot in the pads is intentional, and not the result of Synthopia 2 defect (if any such was in the recording, you would have spotted it clearly).
You may listen to DSK's version here.


VSTi
Year
Bits
Rating
Role A
Synthopia 2
2009
32
Fair
Role B
ChoirZ
2010
32
Fair
Role C
Ethereal PadZ
2008
32
Fair
Role D
ZitH
2006
32
Good
Texture
Darkness Theory
2008
32
Good

 

EFM

EFM's synths worked like a charm. A developer with long history in the free synths market, has released both emulations of virtual analog synthesizers of the 70's-80's and their own synths. A mix of them therefore was chosen for "Amber".  Role A was given to the flagship semi-modular Micromodular, a task that it was proven easy (while not the easiest) to accomplish, while the rich analog Superwave took Role B for which it had no problem. Role C was assigned to Jup-6, an emulation of Roland Jupiter-6, which managed to do very well, with only an extra reverb needed. The tiny Alpha-II took role D, due to its arpeggio functionality, and it was shown more than capable (although Sync to host worked for a constant rate only, and I had to self-tune the arpeggios rate, ending to adding a more human feel to the song). The Texture was given to RJU-60, an emulation of Roland Juno-60, and it was much easy to do, since a nice setting of LPF and HPF, with some discreet S&H gave quickly the desired "water" timbre. The CPU remained between 40%-45%, which I consider a good performance, given the complexity of Micromodular and Hyperwave. Program changes worked like a charm (although at the final recording there exists one slight spike at a change). Overall, my experience with EFM was very satisfying. Their synths (and there are many to choose from EFM) maybe lack some fancy GUI (if someone cares about this), but some of them are more than capable and the worklfow is straight and easy to adapt. The final song version hides its purely analog origin at some extent. The rating for effectiveness amounts to a 2.5 and for efficiency is a 4.5 (just for the FX added).
You may listen to EFM's version here.


VSTi
Year
Bits
Rating
Role A
Micromodular
2006
32
VeryGood
Role B
Hyperwave
2007
32
VeryGood
Role C
JUP-6
2003
32
Good
Role D
Alpha-II
2006
32
Fair
Texture
RJU-60
2003
32
Good

 

Electrostudio

Here's a label that specializes in emulating analog 70's synthesizers. As expected then, they produced the most analog-like version of "Amber". My selection was driven by the specifications of each synthesizer in terms of polyphony, since the emulations include the restrictions of the early hardware. Therefore, polyphonic SixMonthJune, an emulation of Roland Juno-6 took Role C, and polyphonic Micromoon, an emulation of Moog Micromoog, was assigned Role A. Both of them provided nice sounds and worked like the charm. The same happenned with the rest synths, the esLine, an emulator of Arp Solina, which took Role B, and the duophonic Or2v, an emulation or Oberheim Two Voice. The Texture was given to the duophonic MoonSonoSix, an emulation of Moog Sonic Six, and it was much easy to accomplish. No extra FX was needed. The CPU never exceeded 40%, verifying that these excellent emulations are also efficient ones, featuring a commercial quality. If you want to hear how the ambient song would sound in 1975, then here it is. For those who are into retro and analog sound and enjoy a realistic GUI, Electrostudio is one of the best choices, if not the best. All technical aspects (program changes, etc.) worked fine, so in regard to the rating it's counted a 2.5 for effectiveness, and a 5.0 for efficiency.
You may listen to Electrostudio's version here.


VSTi
Year
Bits
Rating
Role A
Micromoon
2010
32
VeryGood
Role B
eSLine
2010
32
VeryGood
Role C
SixMonthJune
2009
32
VeryGood
Role D
Or2v
2010
32
Good
Texture
MoonSonoSix
2010
32
Good

 

EVM Synths

EVM Synths has released a variety of VSTis, which are generally reliable and decent. Some of them are also pretty powerful and rich, like Ultrasonique and DvX Vega. For this task, the additive Geko was given the prestiguous Role A, for which it had no problem. The paddy Omnicron took Role B while the more rythmic paddy Role C was given to Ultrasonique. DvX Vega took rightfully Role D and the Texture was assigned to the specialized Placebo. The setup was easy and straightforward. Program changes worked fine, and the CPU remained below 35%, a very satisfying performance. Some synths required an extra Ping pong delay and a Chorus. The rating for effectiveness amounts to a 3.0 and for efficiency a 4.5 (due to the extra FX needed). EVM's synths are efficient and, despite their age, they offer modern, contemporary sound which can provide quick solutions to any project and music genre.
You may listen to EVM Synths' version here.



VSTi
Year
Bits
Rating
Role A
Geko
2001
32
VeryGood
Role B
Omnicron
2001
32
VeryGood
Role C
Ultrasonique
2001
32
Good
Role D
DvX Vega
2003
32
Good
Texture
Placebo
2006
32
VeryGood

 

Full Bucket Music

This is a contemporary provider of free VSTis, either custom ones or emulations of vintage analog synthesizers. For this setup, the custom synths from Full Bucket Music were selected. Role A was given to the formant Scrooo, for which it had no problem. The rich hybrid subtractive Qyooo was assigned Role B, while Role D was given to the standard FM/subtractive Blooo. For the paddy Role C, the Deputy Mark II was chosen, an emulation of vintage string synthesizers. Finally, for the Texture, the virtual analog Ragnarok, another emulation of vintage analog synthesizers was selected. All synths worked like a charm in terms of behavior. I only added reverb to Scrooo and Deputy Mark II, to enhance the ambience a bit. A very sweet version was produced, hiding its analog backbone at a large extent. The CPU was minimal, remaining below 30% at all times. Overall, working with Full Bucket Music synths was one of the most pleasant experiences. Note that this label requires changes on a patch to be saved somehow in disk (e.g. using their own patch format or .fxp), otherwise they are not preserved after program changes. This is supposed to be an add-on, and it reflects a contemporary settings trend that is followed by some other labels too, in order to prevent accidental changes in patches while playing the song. For the effectiveness the rating is a 4.0 (e.g. the vintage strings of Deputy Mark II are much appealing) and for the efficiency it's a 4.5 (due to the extra effects added).
You may listen to Full Bucket Music's version here.



VSTi
Year
Bits
Rating
Role A
Scrooo
2018
64
VeryGood
Role B
Qyooo
2018
64
VeryGood
Role C
Deputy Mark II
2017
64
Good
Role D
Blooo
2018
64
VeryGood
Texture
Ragnarok
2017
64
VeryGood

 

GTG Synths

This is a legacy provider that specializes mainly on analog synths, but they also provide some FM and phase distortion too. Role A was given to the FM synth FM4. The challenging Role B was assigned to MX2008, a hybrid pad-fx synth, Role C was handled by a relative of MX2008, the analog ST2008 while the analog NR2010 undertook Role D. For the Texture, JP7A, an emulation of Roland Jupiter-7 was used. All of them went through the test without any problem. In general, GTG provides reliable and easy-to-work-with synthesizers, and this was also proven here. The result version sounds somewhat analog, due to the very many analog VSTis used, but on the other hand I had not any issue with program change messages or CPU (it was never above 35% confirming GTG is an efficient label), plus NR2010 demonstrated unexpected capabilities. Overall, it's always nice to work with GTG Synths, and although the song was not asking for an analog sound GTG managed to provide a sweet version. For the effectiveness is counted a 2.5 and for the efficiency is a 4.5, due to extra reverb on FM4.
You may listen to GTG Synths' version here.


VSTi
Year
Bits
Rating
Role A
FM4
2008
32
VeryGood
Role B
MX2008
2008
32
VeryGood
Role C
ST2008
2009
32
Good
Role D
NR2010
2010
32
VeryGood
Texture
JP7A
2010
32
VeryGood

 

Krakli

Krakli has provided almost any kind of free VSTis, in a span of almost two decades. The first thing I realized with Krakli's synthesizers is the quality in sound and the versatility of the synths. The second thing was that none of the synthesizers would support any program change. Ever. Furthermore, such an issue was reported to the developer at least a decade ago, however it seems they chose to ignore the fix. So how to deal then with my song, without any program changes support? One option is double instances, which would raise the number of synths to 9 (that was the case, f.ex. with Acrobatics). Another option is to use a multi-timbral synth, to cover some secondary lines. This is exactly what I did in case of Krakli, where the multi-timbral K700 was called to provide support, resulting to needing only 7 synths in total. Given that Krakli's synths were not proven CPU-efficient either, saving 2 additional instances from my setup can be seen as an achievement. Role A was therefore shared between two instances of the additive RGB8, which is a very nice sounding synth by the way. Role B was shared between the aged but capable physical modelling synthesizer StringZ (the original choice) and one timbre from the first instance of the multi-timbral FM synth K700. Role C was shared between the other timbre of the first instance of K700, and one timbre of a second instance of K700. Role D was shared between the pads-fx synth Cygnus-O (the original choice) and the other timbre of the second instance of K700. If you are already confused what is played by what, then don't. Enough to say that K700 (which, besides being multi-timbral, is a powerful synth on its own) saved somehow the day. The Texture was handled by Vurtbox, a synth with an unbelievably counter-intuitive GUI. The final song version was perfect. If I would only know how to stop Vurtbox however. The CPU was mainly around 60% but it was so unstable that it had several 100% spikes (I stopped counting after 5), confirming that "quality costs efficiency" for Krakli synths. For the effectiveness the rating is a 4.5 (the only penalty concerns Vurtbox which denies my Note Off message and stops abruptly at the end of the recording) and for the efficiency it's a 1.5 (e.g. besides the double instances, I count here also my suffering to get my texture from Vurtbox, see workflow-hindering GUI rule).
You may listen to Krakli's version here.


VSTi
Year
Bits
Rating
Role A
RGB8 (A) / RGB8 (B)
2008
32
VeryGood
Role B
StringZ / K700 (A)
2008
32
VeryGood
Role C
K700 (A) / K700 (B)
2009
32
VeryGood
Role D
Cygnus-O / K700 (B)
2010
32
Good
Texture
Vurtbox
2010
32
Poor

 

Kriminal

This is a label with more than a decade history in the free VSTi market, releasing not only analog ones, but also phase distortion and FM synths. There are very many available instruments, so there are many choices and configurations to choose from. For the needs of the song then, Role A was given to the "keyboards-type-sounds" Synth 2, and it was handled very effectively and efficiently. Role B was planned for the additive Twilight, however as an exception to the rest Kriminal VSTis, Twilight was shown very CPU-heavy, producing unforced spikes and also denying program changes. It got substituted therefore by the phase distortion Czar which took the opportunity to show superior performance for the role needs. Another powerful synth, the subtractive SartreXT undertook Role C and excelled in it. Tsunami, yet another analog, featuring 4 Osc this time, was in Role D. The monophonic but very powerful Victor handled the Texture. The CPU remained very low, never above 25%, confirming that Twilight, which was tried but not selected, is an exception, or perhaps a badly designed case in Kriminal's synths. The song version was sweet, however sadly some clicks occurred at some random points in the song, which you can hear in the recording (although the CPU was low). I haven't spotted which VSTi was the cause, but I suspect Victor, due to the complexity of the routing and the stretching it underwent (but keep this only as a speculation please, for now). Kriminal's synths lack some fancy GUI but they are capable and efficient (with some exceptions). Their signal routing is straightforward to grasp (from their GUI), and the effort to learn them is minimal, making Kriminal a quick alternative for many music genres. For the effectiveness Kriminal gets a 3.5 and for the efficiency another 3.5.
You may listen to Kriminal's version here.


VSTi
Year
Bits
Rating
Role A
Synth2
2008
32
VeryGood
Role B
Czar
2007
32
VeryGood
Role C
SartreXT
2007
32
VeryGood
Role D
Tsunami
2007
32
Good
Texture
Victor
2007
32
Good

 

KX77FREE

This label has released several synths, mostly built around a factory-made sound engine by KX77FREE, which as far as I understood, incorporates custom code in addition to Synthedit routines. They are mostly semi-modular, featuring powerful matrices, while some of them have additional features, like sequencers, etc. Some customization in their code allows them to emulate better an analog signal (a feature I found also elsewhere, e.g. in Psychic Modulation synths), by slightly detuning the note. For the needs of this song, I soon realized that program changes support wouldn't be the case. This time it was clear that such functionality was intentionally abandoned in the newer 64bit synths (together with any patch bank handling), since the older 32bit versions were fully  supporting it. I used the 32bit monophonic, seq-synth, KX-Step for Role D, and it handled it successfully. KX-Modulad was at the single-line Texture, also successfully. For the rest roles, double instances were used, resulting into having a total of 8 instances in the mix. Role A was given to KX-PolyM-CSE (2 instances), Role B was assigned to KX-Polymod (2 instances), and Role C went to KX-Synth16 (2 instances). Don't ask me why these synths to each role, because I don't know the answer, since I feel that the three synths could have taken any of the other two roles, being pretty much equivalent into what they achieve and how they achieve it. The sound from KX77FREE was better than average. Just as with Krakli synths however, this came with the cost in CPU. The average CPU remained at least around 60%, with frequent spikes on 100% CPU, given that 8 instances are more than the original plan of 5 instances, and that there was no way to compensate (as for Krakli were I used a multi-timbral synth). Despite the richness of its sound, an extra ping pong delay had to be added for KX-Polymod. The final version is sweet and decent, albeit quite analog-sounding. It's a pity however that program changes are not supported at any newer synth, given that KX77FREE claims, in their webpage, that their synthesizers are designed for use by experienced musicians - wouldn't an experienced musician need program changes? For their effectiveness, KX77FREE score a 3.5, and for their efficiency a 1.0 (see above issues).
You may listen to KX77FREE's version here.


VSTi
Year
Bits
Rating
Role A
KX-Polym-CSE(A)/KX-Polym-CSE(B)
2017
64
Good
Role B
KX-Polymod(A) / KX-Polymod(B)
2017
64
Good
Role C
KX-Synth16(A) / KX-Synth16(B)
2017
64
Good
Role D
KX-Step
2007
32
Good
Texture
KX-Modulad
2017
64
Good

 

L-Day

This label has some history dating back to 2007, but it is also active currently, with few, in total, but chosen synths. For the requirements of the song, I gave Role A to EightySix, which was bragging for its bell piano (but for my needs, this preset had to go under heavy manipulation). Role B was given to the additive Antopya (a KVR-competition synth), which, like most additives, was able to cope with the demands. Athmonova, a supersaw analog VSTi took Role C and added its own spice to it. Role D was assigned to Yonu-60, an emulator or Roland Juno-60, which was handled with good success, since it had the Juno arpeggiator (which didn't sync to host, but it was easy to self-tune it, adding a human touch to this version). Running out of options, the Texture was assigned to Poly-LD8, an emulator of Korg Poly-800, a synth which we were using in my music group in school. Without problem, a kind of water texture was relatively easy for the 80's digital (combining the HPF and the LPF). No additional FX plugins were needed. The CPU remained impressively low, between 20%-25%, and the program change messages worked like a charm. Setting everything up was quick and the result version was very pleasant to my ears. Overall, L-Day appears to be a developer that aims for quality in what they release, and their instruments, whether analog/digital emulations, or the powerful and modern Athmonova and Antopya (each having a special personality), sound nice, can be a useful addition to many projects, and you shouldn't miss them. It's pity for this developer that they have released only 5 synthesizers. Given the quality of them, I would be happy to see more from L-Day. The rating amounts to a 3.5 for effectiveness, and to a 5.0 for efficiency.
You may listen to L-Day's version here.


VSTi
Year
Bits
Rating
Role A
EightySix
2018
32
VeryGood
Role B
Antopya
2009
32
VeryGood
Role C
Athmonova
2010
32
VeryGood
Role D
Yonu-60
2007
32
Good
Texture
Poly-LD8
2018
32
Good

Novaflash

It is a pity that Novaflash hasn't provided us with even more free VSTis. Its available set of synths is a great work. It managed to produce a very nice version of the song, quickly and efficiently. Role A was assigned to the flagship Quack, which performed excellently. Phasm, as a pads synth, took the more 'paddy' Role C, while Ancestor, although officially just an organ, handled Role D - both of them very successfully. Role B was given to the physical modelling Phyzmod, which was shown a bit harder to manipulate, especially when trying the choir pad sound. Noisework took - what else- the Texture side, and had no problems with that. Seriously efficient in terms of CPU (25%-35%) all synths of Novaflash demonstrated professionalism, handling program changes fast and smoothly. Overall, a 3.5 for effectiveness, and a 4.5 for efficiency counts for Novaflash, a set of synths that compete easily professional VSTis. Keep in mind that these synths are very powerful tools. You might want to check also the rest of Novaflash free synths and effects.
You may listen to Novaflash's version here.


VSTi
Year
Bits
Rating
Role A
Quack
2012
32
VeryGood
Role B
Phyzmod
2013
32
Good
Role C
Phasm
2009
32
VeryGood
Role D
Ancestor
2014
32
VeryGood
Texture
Noisework
2009
32
VeryGood

Novakill

A developer with decent and powerful synths for their age, gave its fight for this ambient song. Ultrakiller took Role A, as a more powerful synth, while the phase distortion Pisscutter was given Role B (it also has an interesting Sweeper feature and a manual, by the way). Both of them provided satisfying sounds. Role C was assigned to Brainkiller, the older matrix analog, and NitrousNEO handled Role D, again without having problems in terms of sound design since all of them have the rythmic element (gating). Finally, Novakiller provided the Texture (rare to find a noise source among Novakill synths, but fortunately some have it). The efficiency was not completely satisfying, since although the CPU during live play was mostly around 35%-40%, at a couple of program changes there were audible 100% peaks. Overall, these synths are heavier than others tested (with Ultrakiller a bit heavier than the rest Novakills, and the Nitrous series a bit lighter), but not too heavy to be useful, while it is acknowledged that they provide rich goodies, such as unison, gates, which in my case were ON (unison was at 2 voices and gates+LFOs were running in three synths). My efficiency rating would have been lower, but the easiness I encountered to create the desired sounds compansates greatly. Overall, I consider Novakills synths a good adder to a set, due to their rich design features for many contemporary music styles, and also their straightforward and easy workflow, but perhaps without overdoing it, especially in regard to unison settings (actually, better stay on one needed instance, perhaps two). I also observed that sometimes the program change gets blocked if the control messages are in different DAW tracks, but it works fine if the messages are on the same track (an artifact I haven't seen to other synths). But overall, all synths supported program change messages. I also felt that both Ultrakiller and Pisscutter needed a reverb for their roles, and added one. The efficiency score gets to 3.0 and the effectiveness score is a 3.5 for Novakill (it is not better because I had a a rather sharp audible peak in the recording, due to the simultaneous program change events).
You may listen to Novakill's version here.



VSTi
Year
Bits
Rating
Role A
Ultrakiller
2009
32
VeryGood
Role B
Pisscutter
2009
32
Good
Role C
Brainkiller
2008
32
VeryGood
Role D
NitrousNEO
2010
32
Good
Texture
Novakiller
2008
32
Good

Odosynths

This label has produced numerous FM and virtual analog synths over a decade. For "Amber", I chose Purple, a 4-operators FM, and it handled the lines successfuly. The analog Dreamer came at Role B, when my first option, Double Six, crushed upon opening to my DAW, and when my second option A-Pad wouldn't support program change messages. Role C was given to the arp synth Microwave Oven, to take advantage for a rythmic pad. The Texture was given to Noisie, which is not a noise synth, as the name would imply, but it's a regular (capable) virtual analog. It got punished to Texture line when it showed that it wouldn't support program changes either (original choice was for Role D). Considering that another analog, Noisy -also not a noise synth as the name would imply-, had also crashed (for which however I was able to update the version later, to a stable one), my options from around 40 synths initially (yes, that many) started running low. Seeq-One therefore, a sequencer synth, as the name -finally- implies, handled Role D, and the role's pseudo-arpeggio needs. Some audible residuals in program changes, several-to-many audible clicks in other times, were the little artifacts that disturbed the recording, which however had succeeded in producing a melodic version (I don't know which synth(s) was responsible). Overall, from the song experience, it appears that some Odosynths are capable, and allow for a detailed configuration, but during play they may not be so reliable as others tested here. In addition, they may have compatibility issues (my versions were bridged, I don't know how they perform elsewhere). The sound was satisfying nevertheless, and no extra FX was needed. Odosynths is one of these cases that creates a high expectancy, from when the sounds are ready, only to spoil later the final result with "unforced errors". They are however very efficient VSTis since my CPU remained always below 25%. Odosynths scores a 3.0 for effectiveness (clicks and gaps reduce the listening experience) and 2.5 for efficiency. Did I mention that, despite its age, Purple remains a very good FM synth?
You may listen to Odosynths' version here.



VSTi
Year
Bits
Rating
Role A
Purple
2005
32
VeryGood
Role B
Dreamer
2004
32
Good
Role C
Microwave Oven
2012
32
Good
Role D
Seeq-One
2005
32
Good
Texture
Noisie
2007
32
Fair

Psychic Modulation

Psychic Modulation has produced a series of analog synths were the sequencing/arpeggiator element is almost always existent. Their synths with the time modulation capabilites make an excellent candidate for ambient and electronic music, in general. I did't have any issue to design the sounds. The only issues had to do with compatibility and functioning, but it was still very limited. For Role A, I chose Meteorite-2, a 3-osc analog while Role B was given to Photon-2, a hybrid experimental, when the first choice Metatron crashed upon loading in my DAW (although in the past it had worked in other instances). Role C was given to Subconcious, another hybrid synth, and the arpy Role D was given to Arptron, a phase distortion sequencing synth. Finally, at the Texture was Aethereal, a superb texture synth. None of the choices had the slightest issue in designing the required sounds. Psychic Modulation synths share some common capabilities and their design reveals straightforwardly their signal routing, so the design can be done also quickly. On the other hand, despite their age, they can potentially work as inspiration tools, due to their wide range of abilities. So, where is the catch? There is no catch, at least not a significant one. For example, there were a couple of audible clicks in the recording, I had to reconfigure Photon-2 and Arptron several times because some parameters wouldn't change upon the program change message (like f.ex. the Volume), and Arptron asked for channel 1 in order to receive Midi. The CPU was at most 30%, confirming that Psychic Modulation synths are efficient.The result version is sweet and no extra FX were needed. If someone wants to avoid a wavetable synth, and shape their rythmic or paddy sound as they like, using basic waveforms, these VSTis are an excellent candidate (plus one may use the phase distortion capabilities of Arptron). The effectiveness amounts to 3.5 and the efficiency another 3.5.
You may listen to Psychic Modulation's version here.



VSTi
Year
Bits
Rating
Role A
Meteorite-2
2005
32
VeryGood
Role B
Photon-2
2005
32
Good
Role C
Subconcious
2006
32
VeryGood
Role D
Arptron
2005
32
Good
Texture
Aethereal
2009
32
VeryGood

Saltline/Drzewo

The developer from the UK with the Polish name was proven an excellent candidate. Role A was given to Ichi, which is a superb VA synth. The rich Swierk got Role B, which satisfied with only an added Ping-pong delay, and Role D was given to monophonic Brzoza, which also demonstrated capabilities and efficiency (giving there also a Ping-pong delay, which is fairly ok, since these synths are provided without built-in FX). Shown a little bit harder task, since it is an additive synth, ADHDi was able to cope with Role C, but required more time for tuning. The (more to the experimental side) San took the texture task which fullfilled without any issue. The CPU effort remained between 35%-50% which is a very good performance for the abilities of these synthesizers. Program changes worked fast and smooth. Overall, Saltline's version is rated with 4.0 in effectiveness, and 4.5 in efficiency (e.g. just for the added FX) - a really good, all-round candidate, whom synths can openly compete commercial ones.
You may listen to Saltline's version here.


VSTi
Year
Bits
Rating
Role A
Ichi
2009
32
VeryGood
Role B
Swierk
2015
32
VeryGood
Role C
ADHDi
2014
32
Good
Role D
Brzoza
2014
32
VeryGood
Texture
San
2008
32
Good

Syncersoft

I don't know what I was thinking when I included this label in my test sets. It managed to produce the most cacophonic version, so much that it makes almost impossible to finish listening till the end of the song. Everything started from the fact that upon loading the five synthesizers the CPU was skyrocketed to 50%. That is, without playing any note. After I started editing the sounds, selecting synthesizers and voices, the CPU went to 100% and it was remaining there. Again, without yet playing the song. As you may have imagined, I could hardly attempt to hear some parts here and there, just to ensure some basic mixing levels. Moreover, the instruments sounded completely detuned. I did my best try to bring them into some alignment, alas without much success, since it was not possible to play more than 2 synths simultaneously. Overall, I cannot think of any way these synths could be useful to a musician. Which synth took which role is not important here. It is enough to say that all synths of this group, except maybe Polyvoks, share practically the same (poor) engine, which also happens to be unbelievably CPU-heavy for what it provides. This incredible set of synthesizers produced a version which might be considered by some even as a "cult" version, or just as a silly one, but since it was anything but intended from my side, the rate can only be a 0.0, just to be fair with the other candidates. Regarding efficiency, Syncersoft scores a 0.5 for supporting program change messages. Which, of course, due to the heavy CPU-load it was failing consistently during play, resulting into having unwanted patches at the final recording. Being totally frustrated, I felt so relieved when I went to the next developer.
You may hear Syncersoft's version here. Or you may not.


VSTi
Year
Bits
Rating
Role A
SubsTractor
2012
32
Poor
Role B
Analog Galaxy
2012
32
Poor
Role C
Analog Voice
2012
32
Poor
Role D
Polyvoks
2012
32
Fair
Texture
Analog Nexus
2012
32
Poor

Togu Audio Line

The popular VSTi and FX developer had its own version of "Amber". Despite the fact that three of the four synths were emulators of analog hardware, it managed to provide an atmospheric version, much contemporary, to my understanding. The key was Noisemaker, the flagship of the brand, which took the responsibility of Role A, and the atmospheric piano. However, with Noisemaker came also the bad news, since there was not way to change the patches with messages. Fortunately, the rest ones functioned perfectly well and the number of instances raised only to 6. Not that it would matter the TAL tools, since the total CPU remained between 20%-25%. These are very efficient synthesizers, and it is also possible to get good and complex sounds from the VA ones too, like the U-No-62 (Role B) and the Elek7ro-II (Role C). Actually, I used the Bassline to get the Role D, which is the flute and the pseudo-arpeggio percussion (it was rather easy, since I'm skilled with my hardware SH-101, and despite the fact that the filter of Bassline is anything but the filter of SH-101). For the Texture role, the other analog, Kunz & Knobel, aka RealSynth was planned, but my bridged version wouldn't load to Cubase. So the Texture was given to Elek7ro-II. All in all, it is always fun to program TAL VSTis, but this time I especially enjoyed stretching U-No-62, to get that melancholic violin of Role B. TAL scores 3.0 for effectiveness (for the rather analog sounding), and 3.5 for efficiency (fail to load RealSynth, and double instance for Noisemaker counted here only).
You may listen to TAL's version here.


VSTi
Year
Bits
Rating
Role A
Noisemaker (A)/Noisemaker (B)
2017
64
VeryGood
Role B
U-No-62
2010
64
VeryGood
Role C
Elekt7ro-II
2010
64
VeryGood
Role D
Bassline
2010
64
Good
Texture
Elekt7ro-II
2010
64
VeryGood

U-he

This is a popular label, that has released four free synthesizers recently - each with each own personality, while they have also best-sellers in the commercial market. A lot of sound synthesis experience and design in this developer then. For the paddy Role B, Triple Cheese was the best candidate. More paddy needs for Role C, and it was given to Zebralette. The more rythmic Role C was taken by Podolski. The prestigious Role A was handled by Tyrell Nexus N6. In the Texture, and since only four free VSTis were available, Podolski was again chosen, which offered an extra motion to the water effect. There were no issues during design, perhaps it would take a bit more time to stretch Podolski to flaw water, but the result compensates. There were no issues in sound output either, the CPU remained between 30%-50%, and program changes worked like a charm. Just like other contemporary labels (e.g. Full Bucket Music), U-he requires that you save your patch somehow (e.g. as .fxp or their own preset) in order to preserve it while program changes happen - to know this may save you time. An extra ping pong delay was considered for the analog Tyrell, but the result version is generally sweet, a little bit analog (Tyrell at the piano), and melodic. U-he scores a 4.0 for effectiveness, and a 4.5 for efficiency.
You may listen to U-he's version here.


VSTi
Year
Bits
Rating
Role A
Tyrell Nexus N6
2015
64
VeryGood
Role B
Triple Cheese
2015
64
VeryGood
Role C
Zebralette
2015
64
VeryGood
Role D
Podolski
2015
64
VeryGood
Texture
Podolski
2015
64
VeryGood

Xenobioz

This label has provided several free powerful synthesizers, combining Synthedit modules with custom code for optimization. For the needs of the song, Role A was given to the wavedraw/phase distortion unison synth Lynx, while Dulcet, a physical modelling instrument took Role B. Role C was assigned to the morphing synth Lucid. For Role D, the original choice, the older wavedraw synth Kruudster had to be replaced by the exprerimental morphng synth Funxion, when it was realized that Kruudster wouldn't support program changes. The options were running low, and another phase distortion synth, Lektro (which doesn't feature any noise source), was called to support the Texture role, but it succeded beyond any anticipation. Generally, all synths performed fine. Additional FX were needed, that is ping pong delay+rev. to Lynx, chorus+rev. to Lucid. The synths worked as a charm, the sound output didn't have any defects, and the result version is sweet and intriguing. The CPU remained low, 30%-40%. For effectiveness, the rating is 4.5 and for efficiency is 3.5 (obligatory change of Kruudster, additional FX).
You may listen to Xenobioz' version here.


VSTi
Year
Bits
Rating
Role A
Lynx
2017
32
VeryGood
Role B
Dulcet
2011
32
VeryGood
Role C
Lucid
2014
32
VeryGood
Role D
Funxion
2013
32
VeryGood
Texture
Lektro
2012
32
VeryGood

 

So, where is H.G. Fortune?

For some of you, even by reading here the word "ambient", the first thing that has come to your mind is H.G.Fortune. Indeed, the German label, since very many years has associated its name with the free (or commercial) ambient soft synthesizers, until their founder, Günter Hager, passed away, a few years ago. A name with a legacy, H.G.Fortune has produced numerous VSTis, either free, or commercial (from which many were set as freeware before the end), focusing commonly in ambient sound, pads and textures. So someone might say that their presence in a review for an ambient song like this one, would have been the most natural thing. And I agree. H.G.Fortune should have a presence here. On the other hand however, I consider that it is not my role to judge H.G.Fortune. I myself consider H.G.Fortune synths to act more as pure inspiration tools, rather than regular synths, and perhaps they belong to their own category, not to be compared with anything else, when it comes to ambient sound. To attempt to stretch them to fit regular voices wouldn't have been natural with HGF synths. Therefore, and just for completeness, I produced a complementary version of "Amber", using this time a semi-random synth group pickup from HGF. It's the same song, but at the same time it cannot be the same. 
You may listen to H.G.Fortune's version here.


The Winner


Just like in a Football World Cup, there is also here a winner, and a few others who managed to place themselves at a close position, while some others didn't make it. The results of this review included for me indeed an element of surprise (which was the case with my previous review too), which might point that it is better to try to find out the pros and the cons of each synth/label in a real task, rather than just making quick assumptions based on specificiations/GUI/reputation, etc.

In the following, in case of an equal scoring, the names appear in alphabetical order. In parentheses, it's referred the total score acquired. The top three positions are referred, which can be held by a single label, or in case of a tie it's the labels that satisfy the condition for the position, but no more places after presenting at least three labels are reported.

Top - 3 Overall

 

According to the results of this review, these are the "all-around" combiners of having both capabilities and allowing for the best flexibility, while providing smooth cooperation, the best of the best.

The first place is shared by four labels:
  • Full Bucket Music (8.5/10)
  • L-Day (8.5/10)
  • Saltline/Drzewo (8.5/10)
  • U-he (8.5/10)

Top - 3 Effectiveness only

 

This review for "Amber" has shown the following "fine-shapers". These are useful labels were the best possible sound quality is wanted, while efficiency is not necessarily of concern.

The first place is shared by two labels:
  • Krakli (4.5/5)
  • Xenobioz (4.5/5)
The third place is shared by four labels:
  • Big Tick (4/5)
  • Full Bucket Music (4/5)
  • Saltline/Drzewo (4/5)
  • U-he (4/5)

Note: One of the Big Tick synths comes for free with a magazine, in a 1.5Gb-bundle that includes VSTis, for $5.00, e.g. it's not completely free.


Top -3 Efficiency only

 

From the conclusions of this review, these are the "workhorses". They may need some additional tweaking or accepting some compromises in sound, but they work like a charm and allow easily for large projects to become realized.

The first place is shared by two labels:
  • Electrostudio (5/5)
  • L-Day (5/5)
The third place is shared by seven labels:
  • EFM (4.5/5)
  • EVM Synths (4.5/5)
  • Full Bucket Music (4.5/5)
  • GTG Synths (4.5/5)
  • Novaflash (4.5/5)
  • Saltline/Drzewo (4.5/5)
  • U-he (4.5/5)

Note: The difference between the first and the third place, at this level, was in all cases the additional use of one or more FX. No other technical issues were encountered to the labels in the third place, as compared to the ones that are in the first place.


Noisy Raspberry award



Just like in any decent competition, there is here a special Noisy Raspberry award.
This is awarded to the following label:

  • Syncersoft (0.5/5)


Congratulations to all! 



You may also want to know the statistics of the rating. The average score for effectiveness was 3.22 and for efficiency was 3.28. The standard deviation of effectiveness was 0.91 and for efficiency was 1.28. The (Pearson) correlation between effectiveness and efficiency was 0.22 which is low (e.g. mostly non-correlated), in other words, there are not many all-around good or bad companies, neither are there companies that sacrifice the one quality for the other one (that situation would carry strong negative correlation), but how good is a company in one quality is not generally an index to how good is for the other one, because only one in five companies is persistent (good or bad) in both qualities. It's a mixed landscape, or rather, a mixed soundscape.


Epilogue 



As an epilogue, I can testify that in this short trip that I did into each and every label, I felt like being in some kind of car test-drive. Sometimes I had a Rolls-Royce to drive, and I could lay back and enjoy the experience. Other times I had a Wrangler Jeep, that I needed to tame in order to get to the desired goal. And, of course, there were cases where my car broke down even before the test-drive, and I had to make autostop to get back home. After doing this test, I can say that I feel wiser regarding the free VSTi market. And by sharing this experience, I hope to help the reader/listener too, and perhaps save time from his/her own search.



About me


I'm a music hobbyist enjoying writing and instrumenting music in my spare time. My involvement with the music begins since 1985. Among others, I bought my Roland SH-101 in 1986, my Yamaha TG-500 in 1992 and my Yamaha Clavinova and Yamaha SY-35 in 1994. I still have (and use) the first three ones, together with my Access Virus TI2, Clavia Nord Lead A1, Roland SH-201, Novation Peak, Behringer DeepMind 12, Elektron Digitakt, Roland JV-1080, Behringer Neutron, Behringer Model D, and my preferred software (VST) instruments, controlled by Samson Graphite 49, E-Mu X49 and Omnitronic FAD-9 (my SH-101 is controlled using Doepfer Dark Link).