Ableton Forum

hi guys!

so i seen the video of the workshop from robert henke and he shows how the hi quality mode of operator changed the quantization artifacts and the overall quality of the sound from operator, so my idea was to see how good is the quality of other softsynths! and i think the results are very interesting and are exactly what i was hearing all the time!
with all synths i made a simple sine wave at C3 with circa -0.2 db peaking and because some synths don´t reach this level, i placed the live utility plugin after every synth to level it exactly and to give a fair result.and then i searched the highest peak in the highfrequency area where all this dirt exist;-)

and here are the results:
1. Absynth4 -111 db (absolutely the winner;-))

2.Acces virus TI Wavetablemode over USB16bit -106db (ok its not really a softsynth but i tested it

3.MASSIVE in Ultramode!!! -104db

4.Zebra 2- with normal OSC and FM OSC -103db

5.FM8 in Highresolution mode -101db

6.Reaktor 5- Carbon2 -100db

7.MASSIVE in Ecomode -97db

8.Gladiator in Highendmode -80,3db

9.Rob Papen Predator with 16x oversampling -75,6 db

10.Rob Papen Predator with 1x Oversampling -70,1db

11.Gladiator Normalmode -67,5db

12.Gladiator in Lowmode -28,6db

so these tests were all made with 44.1 khz and inside live, only the utility plugin were between the synts and the spectrum plugin of live. and these tests are not superexactly and i dont say that softsynths with bad results are sounding bad, because the soundasthetic is a personal taste;-) so not everybody like the sound of absynth or the access virus TI! but these are the results and maybe you can add some more synths!

GREETZ

and sorry for my weak english

Call me daft, but what does this tell you?

Lo-Fi Massahkah wrote:Call me daft, but what does this tell you?

a sine wave, a pure sine wave is just 1 frequency, at C3 there should only be a single tone at 130.8Hz. but with the math that's in all these things it's not ideal and other frequencies can be introduced. it seems that the OP looked for the highest artifact (left over junk created by bad math algorithms) and measured how high it was.

-60dB is about where I draw the line for what's noticable, YMMV. anything below that is kinda negligible. 0dB is loud, -100dB is quiet.


andy - interesting stuff!!! thanks!

not all synths use pure sine waves, for example, Operator has like 4 different sine waves, only 1 is pure, the others have different levels of harmonics added in to fatten them up.

did you use Spectrum to check this out?

good work!

Thanks. I see.

it was just for fun and i was a little bit bored today so i tested it, but maybe some people want to buy in the next time a new synth and search for a really clean sound and when they read my test they know what to do! because all software companys writing in their specs "absolutely aliasing free OSC, billion times oversampling " and so on but you see that its not true!

Tone Deft write: "not all synths use pure sine waves, for example, Operator has like 4 different sine waves, only 1 is pure, the others have different levels of harmonics added in to fatten them up."

yep, thats right and a lot of other synths do this to make the sound fatter(its a little bit faky i think ).
when i think right only absynth and the virus (mostly the virus) making the purest sinewave with only 1 harmonic! and they still sound fat!

You should repeat same test with sinewaves of hardware synthesiers. you will be surprised what you'll get.
Also interesting is not only the level of distortion but also how exactly the distribution is. Like, you might have a harmonic overtone that is linked to the pitch of the note you play and the result sounds quite nice. Or you might have some static or non linear related distortion that sounds not as good.
Also: try to sing / whisle a sinewave and post result

Robert

and yes i used spectrum with 16384 FFT-Blocks

Robert Henke wrote:Also: try to sing / whisle a sinewave and post result

Robert

it's pretty easy with that Live vocoder, isn't it?

sylenth - highest peak -98.8 @ 20.9kHz

operator is so clean! -142 @ 20.7 khz


however, even with headphones, I couldn't really tell the difference between the two

Robert Henke wrote:You should repeat same test with sinewaves of hardware synthesiers. you will be surprised what you'll get.
Also interesting is not only the level of distortion but also how exactly the distribution is. Like, you might have a harmonic overtone that is linked to the pitch of the note you play and the result sounds quite nice. Or you might have some static or non linear related distortion that sounds not as good.
Also: try to sing / whisle a sinewave and post result

Robert

oh i dont say its not sounding nice i really like rob papen predator and the sound of it and i never thinked that it will get so bad results (for a softsynth). tomorrow i will test my radikal technologies spectralis and today i´ve also tested the analog outs of my virus and it was at -101db but i think these are not so exactly results because the influence of cables and A/D converters are bigger, or not?

andy hunt wrote:when i think right only absynth and the virus (mostly the virus) making the purest sinewave with only 1 harmonic! and they still sound fat!

that's an interesting post. does a pure sine wave sound fat? should it? I've never heard of a way to quantify 'fat'. IMO a pure sine wave sounds thin. not to argue, it's an interesting point you implied.

Tone Deft you are absolutely right i think its a little bit more complicated and normally i hate people when they use the word "fat" for music or a synthesizer this is all personal taste!

My test with gladiator:

init patch, select WV Sine. Mute everything but osc1/2, turn up master volume till i got it around -0.3, then used with utility. High-end mode selected.
44.1 khz sample rate.

I get peaks at round -83, using spectrum with the same setting. why did you only test low and normal? Even when playing around i never have it on another thing than high.