Stretch/sample rate convert to fix timing - deleterious to quality? (Reaper)

[Dan B]

Hi,

I have some tracks where the performance is fine, save for some shonky timing. In Reaper, you can stretch items to adjust the timing (not pitch), which I understand (though may be wrong) it does via sample rate conversion. (e.g. if you lengthen something, it will show e.g. 1.131543; shorten, say 0.882135 - in each case indicating the rate/ratio at which it is playing back the original).

This is, as I understand it, fairly standard industry standard practice. Whilst I wouldn't really hesitate to do this in a busy mix, the tracks in question are exposed, solo instrumental passages (recorded at 24/96). I'm not particularly noticing any deterioration (though haven't tested), but is this a kosher approach or am I missing something.

Many thanks,

D

[James Perrett]

The process is far more complex than just slowing down or speeding up the sample rate as, if you keep the pitch the same, Reaper has to add or remove sections of audio to match the time added/removed. It uses various pitch shift algorithms to do this and I'll often try different algorithms if I think that I'm hearing artefacts.

You may find more information on the various algorithms Reaper uses at

https://products.zplane.de/elastique-pitch-2
https://www.surina.net/soundtouch/

[Dan B]

Ah yes, of course!
Had I engaged brain, it's obvious that some sort of pitch/time stretch must be required, since changing the sample rate also changes pitch.
Thanks for the links - will check them out!

[hobbyist]

Hi,

I have some tracks where the performance is fine, save for some shonky timing. In Reaper, you can stretch items to adjust the timing (not pitch), which I understand (though may be wrong) it does via sample rate conversion. (e.g. if you lengthen something, it will show e.g. 1.131543; shorten, say 0.882135 - in each case indicating the rate/ratio at which it is playing back the original).

This is, as I understand it, fairly standard industry standard practice. Whilst I wouldn't really hesitate to do this in a busy mix, the tracks in question are exposed, solo instrumental passages (recorded at 24/96). I'm not particularly noticing any deterioration (though haven't tested), but is this a kosher approach or am I missing something.

Many thanks,

D

You may or may not hear it but it is definitely not as good as it was without the diddling, although the sound itself would still be off just better recording of the off tempo sound.

After diddling, the sound will be in tempo, and the pitch the same but there will be some lessening of quality of the sound of that better performance.

[Dan B]

That's what I feared and expected. Re-recording in time is always going to be preferable but not always an option. I guess my question is the extent to which (on sensitive solo instrument performances) the extent to which that's a problem (when making changes of up to say 10 - 20%) ... or common practice?

[Mike Stranks]

That's what I feared and expected. Re-recording in time is always going to be preferable but not always an option. I guess my question is the extent to which (on sensitive solo instrument performances) the extent to which that's a problem (when making changes of up to say 10 - 20%) ... or common practice?

I'm not doing the same thing as you: adjusting one track in a multitrack, recording, but I'm time-stretching tracks - up and down - all the time. Doing it just last night in fact to fit music tracks to video...

My rule of thumb is that as long as you're using something with a good algorithm - and there are various algorithms to do this, often three or four in the one utility - then up to about 5% is usually acceptable. 5-10% can be problematic and anything above 10 is a no-no.

That's me; YMMV

Hope you get it sorted.

[N i g e l]

then up to about 5% is usually acceptable. 5-10% can be problematic and anything above 10 is a no-no.

I would agree that 10% is asking a lot based on my use of Rolands V-vocal which I use to adjust 1 track at a time in a multitrack.

My edited track will be the same length as the original but the position of each syllable can be moved around like it has been recorded on elastic.

Small changes to pull a syllable onto the beat are acceptable but it doesnt take much to get into "special effects" territory. 10%-20% is a lot.

For aligning arbitary vocal to the beat [eg like a radio news item] I would 1st slice up the words, remove dead space, time align the bits and then bounce back into a clip before V-vocaling.

[James Perrett]

You may or may not hear it but it is definitely not as good as it was without the diddling, although the sound itself would still be off just better recording of the off tempo sound.

After diddling, the sound will be in tempo, and the pitch the same but there will be some lessening of quality of the sound of that better performance.

While the sound will theoretically be degraded, in practice modern algorithms are extremely good and the degradation is often inaudible. I often use stretching in Reaper instead of RX's spectral repair to remove blemishes as Reaper's stretching is more transparent and less audible than RX's.

[MOF]

Whilst I wouldn't really hesitate to do this in a busy mix, the tracks in question are exposed, solo instrumental passages (recorded at 24/96). I'm not particularly noticing any deterioration (though haven't tested), but is this a kosher approach or am I missing something.

Am I missing something too, you say you’ve tried it and haven’t noticed any deterioration, so it’s worked?
Or are you saying it’s still not right timing wise? If so then I suggest you use the “elastic audio” facility that Logic, Melodyne etc have to slide individual transients into position.

[Kwackman]

I have some tracks where the performance is fine, save for some shonky timing.

Depending on how much "some" means, it might be better to edit the audio (cutting, moving pasting) to correct the out of time note(s)? There would not be any change in audio quality.

I do this a lot with my, ahem, "vaguely sync" guitar playing I zoom in and slice the offending chord and slip it to a more musical position.

[CS70]

So long the clips you stretch are short enough, nobody'll be the wiser. For vocals, it's a good idea to cut down to individual letters if possible. For other instruments, it's even easier.

It's important that you re-listen to the part you've fixed starting a bit before - like with drums, you can't really get a sense of the rhythm unless you ramp up a few bars..