[mma] - Overdrive - Oversampling broken

Bunky Freaks · Post by **Bunky Freaks** » Thu Feb 05, 2009 2:52 pm

in case the new overdrive effect is oversampled it is clearly faulty. This simple test (project samplerate 44 kHz) shows no oversampling is going on. A sine sweeps up to 22 Khz and processed through overdrive:

http://www.torpedotrickser.de/live8beta ... broken.zip

Once the highest generated harmonic hits nyquist (22050 Hz) it is instantly mirrored back into the audible spectrum without being attenuated.

/edit: removed suggestion/feature wish

Angstrom · Post by **Angstrom** » Thu Feb 05, 2009 3:49 pm

The key point in relation to this issue is that the current implementation is not working as the manufacturer intended.

Advice on what you would prefer (in terms of oversampling amount) should probably be kept separate from a thread alerting them to the existing oversampling being broken. Otherwise your report will get misclassified.

Just my POV, probably wrong.

Bunky Freaks · Post by **Bunky Freaks** » Thu Feb 05, 2009 4:06 pm

true. removed fr. where should requests for live 8 be posted, in "feature wishlist"?

ilia · Post by **ilia** » Thu Feb 05, 2009 7:15 pm

the project's sample rate was >44.1, right?

Angstrom · Post by **Angstrom** » Thu Feb 05, 2009 8:22 pm

ilia wrote:the project's sample rate was >44.1, right?

oversampling within the devices in realtime audio works at multiplications of the soundcard operating frequency, not the sampling frequency of the recorded waves.
Try the example file - with your soundcard at 44.1, and see if the device created harmonics reflect back down when they hit 22khz.
they will

the recorded wave in this instance is not the issue - the additional realtime harmonics generated should be able to rise invisibly (inaudibly) above 22khz. They do not.

ilia · Post by **ilia** » Thu Feb 05, 2009 10:43 pm

Angstrom wrote:oversampling within the devices in realtime audio works at multiplications of the soundcard operating frequency, not the sampling frequency of the recorded waves.

correct

Try the example file - with your soundcard at 44.1, and see if the device created harmonics reflect back down when they hit 22khz.
they will

yes, but it could -- and in fact does, as I just checked -- happen when the project's sampling rate is higher, so this is not an oversampling issue per se (if it is indeed oversampled). And -- any practical oversampling of an overdrive will probably still result in harmonics reflecting back, since the overdrive algorithm will likely generate harmonic content above half the oversampling rate. Anyway, it's for Ableton to decide whether they want to filter those -- I personally am happy with the sound of aliasing...

ilia · Post by **ilia** » Thu Feb 05, 2009 11:11 pm

Also -- I just listened to (and looked at) the swept sine sample without overdrive played back at 192kHz, and there is a whole lot of aliased lower frequency content added that wasn't there at 44.1. Not sure what this means, but it's kind of more disturbing than the original overdrive issue...

Bunky Freaks · Post by **Bunky Freaks** » Thu Feb 05, 2009 11:31 pm

ilia wrote:Also -- I just listened to (and looked at) the swept sine sample without overdrive played back at 192kHz, and there is a whole lot of aliased lower frequency content added that wasn't there at 44.1. Not sure what this means, but it's kind of more disturbing than the original overdrive issue...

that might be due to the resampling of the audio file. For a clean sine sweep at any project sample rate use the test tone plugin from this freeware effects bundle:
http://mda.smartelectronix.com/

Bunky Freaks · Post by **Bunky Freaks** » Thu Feb 05, 2009 11:45 pm

ilia wrote: I personally am happy with the sound of aliasing...

well, this is about a potentially broken implementation of oversampling in overdrive. if it does not work as expected (which i think i found prove for) it should be fixed.

ilia · Post by **ilia** » Thu Feb 05, 2009 11:49 pm

Bunky Freaks wrote:
ilia wrote: I personally am happy with the sound of aliasing...
well, this is about a potentially broken implementation of oversampling in overdrive. if it does not work as expected (which i think i found prove for) it should be fixed.

true, but what's the expected behavior? I didn't even find any mention of it being oversampled...

ilia · Post by **ilia** » Thu Feb 05, 2009 11:54 pm

Bunky Freaks wrote:that might be due to the resampling of the audio file.

if resampling audio at higher rate in Live produces so many artifacts, that's worrisome. I rarely record at >44.1, so never noticed this before.

For a clean sine sweep at any project sample rate use the test tone plugin from this freeware effects bundle:
http://mda.smartelectronix.com/

thanks for the link. just tried it with a tone fixed at 18k (can still hear it -- just barely!), sampling rate 192K, applying overdrive with a high drive and/or high dynamics produces a lot of low frequency content.

Bunky Freaks · Post by **Bunky Freaks** » Fri Feb 06, 2009 5:20 pm

if you think something is wrong that's not related to the original topic and worth reporting please open another thread.

about the oversampling issue, maybe someone from ableton could shed some light on this?

Angstrom · Post by **Angstrom** » Fri Feb 06, 2009 5:58 pm

@bunky

the thread prefix [MMA] has appeared. That means Matthias Mayrock is investigating it.

Once it gets identified as a bug it will usually get assigned a bug ID number, hence the threads with [number] prefixing the user's title.
It may be that Matthias will just fix this issue if he sees a problem, as the Overdrive was written by him, I think.

Amaury · Post by **Amaury** » Fri Feb 06, 2009 6:24 pm

Hi,

We could identify some issue, and fix it internally. It should make it in one of the next betas. Thanks for your attention.

Regards,
Amaury