Overdubbing, Audio to Buffer and Latency

[billw58]

Try this experiment. In Preferences > Recording set Overdub ON, Audio to Buffer to 100 and Latency Correction to 0. Generate a track of a known length (Click or Tone work well). Select all of the generated track and click Record - a new "overdub" track is recorded. Double-click on the new track and note the length - it's a little longer than the original track. This is good, since if the Latency Correction was non-zero the overdubbed track would be pushed back by the Latency Correction amount.

The potential problem is this: the length of time added to the end of the overdubbed track is determined from the Audio to Buffer setting, not from the Latency Correction setting. If you go through setting the latency correction for your system - http://manual.audacityteam.org/index.ph ... tency_Test - and then try overdubbing as described above, the ends of the original and overdubbed tracks will never line up.

"What's this issue?" you may say. "You can get around this by not selecting the original track and just recording from the beginning and letting the recording go past the end of the original track." True. By why should there need to be a workaround? Why can't Audacity add the latency correction amount to the end of the new recording? This would seem to make more sense.

-- Bill

[steve]

the length of time added to the end of the overdubbed track is determined from the Audio to Buffer setting, not from the Latency Correction setting.

I'm not a programmer, but I would imagine that what is happening is that Audacity gets to the end of the selection and stops recording new data, but it has an audio "buffer" that still holds a bit. At this point it can write it to the track or throw it away, and the former is probably better in most cases than the latter as it can always be trimmed off later if required.

[billw58]

it has an audio "buffer" that still holds a bit. At this point it can write it to the track or throw it away

Steve:
It's more complicated than that, unfortunately. The amount of time added is less than the Audio to Buffer setting, and is always less than the latency correction setting. So if you a) set a latency correction amount, b) select the region you want to add the overdub to, and c) click Record ... recording will effectively stop a little before the audio region you selected in (b). You will hear all the way to the end of the region you selected, but you can't record your overdub all the way to the end. You'll only miss a few tens of milliseconds which doesn't sound like much unless you're trying to do a punch-in. Now, Audacity doesn't have pre-roll and post-roll for doing punch-ins so this point may be moot - anyone trying to do punch-ins would quickly learn that you shouldn't select the punch-in region but should instead just click a few seconds before the punch in point and click Record (on a new track, of course). Then edit the punch-in to fit the hole.

Anyway, all this business of punch-ins aside, I still think Audacity should add time equal to the latency correction to the end of the newly-recorded track during overdubbing, rather than an arbitrary amount based on the Audio to Buffer setting.

-- Bill

[steve]

You will hear all the way to the end of the region you selected

I've just tested that, and playback (on this computer) is ending just before the end of the selection.

I see your point about it being slightly imprecise, but since Audacity does not do drop-in recording into the same track*, are there any situations where it matters?

*(making drop-in recordings onto a new track is a better solution in many ways to dropping-in on the original track)