Synths do not always sound what the interface claims

[Dúnedain]

Just found out some synths I frequently use do not do/sound as the interface claims. Probably doesn't matter to everybody, but if you are into sound-design then it definitely matters. For this post it is about volume ADSR envelopes only.

SoundBlaster:

• I set an attack of 1000 ms. I measure it and it is 927 ms (linear). A difference but no big deal;
• I set a release of 1000 ms. I measure 419 ms (exponential). This is just a lot.
• I set an LFO rate of 1.00 Hz. I measure 1.05 Hz.

ES2:

• I need an exponential release of about 300 ms - 320 ms. To achieve this I have to set the release at ES2 at 190 ms.

These are not some free SynthEdit synths. These are synths by respected brands. To me this really comes as a big surprise.

(all under the assumption the wave editors do show the right values of course)

[chk071]

Actually, i noticed the same. For example, on Retrologue, the envelope stage times don't seem to match the length of the actual stages. On the other hand, one has to be careful with that, because some envelope stages are linear, some are exponential, or logarithmic. So, just because there doesn't SEEM to be a signal, doesn't mean there is no signal.

[exmatproton]

Just found out some synths I frequently use do not do/sound as the interface claims. Probably doesn't matter to everybody, but if you are into sound-design then it definitely matters. For this post it is about volume ADSR envelopes only.

SoundBlaster:

• I set an attack of 1000 ms. I measure it and it is 927 ms (linear). A difference but no big deal;
• I set a release of 1000 ms. I measure 419 ms (exponential). This is just a lot.

ES2:

• I need an exponential release of about 300 ms - 320 ms. To achieve this I have to set the release at ES2 at 190 ms.

These are not some free SynthEdit synths. These are synths by respected brands. To me this really comes as a big surprise.

(all under the assumption the wave editors do show the right values of course)

Yep..it is there. We know...and now?? I will still using those synths for sound designing.

[whyterabbyt]

Probably doesn't matter to everybody, but if you are into sound-design then it definitely matters.

Well, if you do sound design by parameter values then it matters. If you do sound design by listening, then no, not so much. Not all synths have caibrated parameter values, its never stopped folk from doing sound design.

[Dúnedain]

Probably doesn't matter to everybody, but if you are into sound-design then it definitely matters.

Well, if you do sound design by parameter values then it matters. If you do sound design by listening, then no, not so much.

If you do sound design by listening you will not always get the desired result. Sometimes you need to have highly accurate values.

[Dúnedain]

Just found out some synths I frequently use do not do/sound as the interface claims. Probably doesn't matter to everybody, but if you are into sound-design then it definitely matters. For this post it is about volume ADSR envelopes only.

SoundBlaster:

• I set an attack of 1000 ms. I measure it and it is 927 ms (linear). A difference but no big deal;
• I set a release of 1000 ms. I measure 419 ms (exponential). This is just a lot.

ES2:

• I need an exponential release of about 300 ms - 320 ms. To achieve this I have to set the release at ES2 at 190 ms.

These are not some free SynthEdit synths. These are synths by respected brands. To me this really comes as a big surprise.

(all under the assumption the wave editors do show the right values of course)

Yep..it is there. We know...and now??

You knew it? Well it is new to me. I am surprised I have to set 190 ms release in ES2 to get a release of 320 ms.

I will still using those synths for sound designing.

[whyterabbyt]

If you do sound design by listening you will not always get the desired result. Sometimes you need to have highly accurate values.

Which times would those be, then?

[aciddose]

For a logarithmic (it is most often not exponential vs. time) curve you have something called an asymptote.

Asymptote: a line that continually approaches a given curve but does not meet it at any finite distance. (A line/point/value which would be reached by a curve/equation if that curve/equation were computed for the input of positive or negative infinity. In other words if the curve continued on to infinity or for an infinite length of time.)

Read the Xhip manual section on envelope timing for a description.

http://xhip.net/synth/

Envelopes
The envelopes are standard ADSR envelopes much like those found in numerous other synthesizers.

Envelope timing
Timings are displayed based upon the time it takes to travel 99% of the step size. For example during the release stage of the envelope the peak of the envelope is 1.0 or 0 dB and the envelope decays toward 0.0 or -inf dB.

The time displayed is the length of time during which the envelope will have reached within 1% of the destination. That is 0.01 or -40 dB.

This may make the time displayed for envelopes seem unusually high in comparison to other synthesizers which use a measure based upon a smaller segment of the curve.

Attack stage asymptote
Envelopes are often described referring to the attack stage as linear. This is often incorrect.

In reality the attack stage is generally the same curve as the decay and release stages however the end point of the curve (called its asymptote) is beyond the peak level of the envelope. In other words the curve is clipped off at some point and the decay begins.

If the envelope were allowed to continue to move toward the asymptote it would only reach that point after an infinite amount of time. This is disadvantageous as we likely do not have an infinite amount of time to wait for the decay stage!

Common asymptotes are 101% to 200% of the peak. This makes the curve appear more linear as varied amounts are clipped off. The envelopes use an asymptote set at 200% which is slightly more linear than some synthesizers although not entirely linear. The decay stage starts at 50% between zero and the asymptote.

The most common timings are based upon an old standard where the time to travel from 10% to 90% is used. Another common timing is based upon a misinterpretation of this standard and is from 0% to 90%. Due to the fact this standard is entirely arbitrary, any step size could be used and remain 100% accurate by its own standard.

[Dúnedain]

Thanks aciddose

[exmatproton]

Just found out some synths I frequently use do not do/sound as the interface claims. Probably doesn't matter to everybody, but if you are into sound-design then it definitely matters. For this post it is about volume ADSR envelopes only.

SoundBlaster:

• I set an attack of 1000 ms. I measure it and it is 927 ms (linear). A difference but no big deal;
• I set a release of 1000 ms. I measure 419 ms (exponential). This is just a lot.

ES2:

• I need an exponential release of about 300 ms - 320 ms. To achieve this I have to set the release at ES2 at 190 ms.

These are not some free SynthEdit synths. These are synths by respected brands. To me this really comes as a big surprise.

(all under the assumption the wave editors do show the right values of course)

Yep..it is there. We know...and now??

You knew it? Well it is new to me. I am surprised I have to set 190 ms release in ES2 to get a release of 320 ms.

I will still using those synths for sound designing.

Yeah, i knew that. And on most occasions, no, i don't care. I care about sound and timing (related to my clock speed), not about timing issues in the 'seconds or milliseconds' scale. I only do accurate stuff when i want to link envelopes or lfo's to my clockspeed. (Milli)second-stuff i do by listening to the sound alone.

If a synth is saying it's doing 1/4, but is actually doing 3/4 or 1/4 dotted for instance, i will write a bugnote to the dev.

[fluffy_little_something]

There are few parameters where I want to know the exact value, tuning for instance. Other than that I just trust my ears. I mean, if the release is too short it is too short, no matter how many ms or s it lasts

Many hardware synths didn't have displays, either. And from what I heard some of those synths were pretty off

[exmatproton]

There are few parameters where I want to know the exact value, tuning for instance. Other than that I just trust my ears. I mean, if the release is too short it is too short, no matter how many ms or s it lasts

Many hardware synths didn't have displays, either. And from what I heard some of those synths were pretty off

+1

[Dúnedain]

There are few parameters where I want to know the exact value, tuning for instance. Other than that I just trust my ears. I mean, if the release is too short it is too short, no matter how many ms or s it lasts

Many hardware synths didn't have displays, either. And from what I heard some of those synths were pretty off

+1

-1

What if I want to recreate the sound of a hardware synth with ES2? Then I need to know exact values so I have to trust the synth I use to recreate the sound. I am not talking about an approximation but about an exact recreation.

Not so strange isn't it

[fluffy_little_something]

-1

What if I want to recreate the sound of a hardware synth with ES2? Then I need to know exact values so I have to trust the synth I use to recreate the sound. I am not talking about an approximation but about an exact recreation.

Not so strange isn't it

Unless you buy a very faithful emulation (maybe even with sysex import), you can forget about achieving exactly the same sound on a different synth, anyway.
And you might get very close despite using clearly different values

[db3]

There are few parameters where I want to know the exact value, tuning for instance. Other than that I just trust my ears. I mean, if the release is too short it is too short, no matter how many ms or s it lasts

Many hardware synths didn't have displays, either. And from what I heard some of those synths were pretty off

+1

+100