I run into this using 2.1.1 on an i7 assus G78 with 8 or 16 gb. I have not yet tried it on my 8-core 16gb desktop deathstar.
I play a backing track in Audacity and record at the same time with overdub & play-through checked in prefs. There’s a few seconds of silence at the start of the BT, before a cue pip, so that I can get ready before ‘hitting it’. THe recording starts as soon as I click the record button. I do not want to mix with the BT at this time. All goes fine but when it comes to playing back the two together it sounds out of step, the cure is to inject a 1/10-1/5 second silence in the leader of the just recorded track. I really don’t understand how this can be, I would expect maybe the opposite because of reaction time or something similar. What I’m listening to is the computer headphone line routed through a wireless headset. The track being recorded is guitar output routed through a Boss-Me80 and a mixer. If anything I would expect the guitar output to drag, not the BT replay
Which brings me to a supplementary, is there a way to mark a backing track so that recording would start when that spot is reached during playback? Or, an alternative woud be to copy a cue pip on the BT exactly over to the one recording whet that spot is reached?
Finally, when using rosegarden I do everything through jack, would that improve on the initial problem stated at the beginning (I don’t use jack with Audacity alone as I don’t see whaty use it would be)?
If you sing or perform in perfect time to the backing track but the stacked performance tracks are out of step, then the Recording Latency adjustment is wrong.
One of the ways to set this is actually jam your headphones next to the microphone and record the backing track.
The illustrations are before and after adjustments. Tick tracks are good because they have sharp waveforms.
If your computer can’t keep up with two real-time live performances, one recording and one playing back, then the latency may shift around and you’ll never hit it pass after pass.
You may not be able to hear yourself during the performance. It’s not unusual for your live voice to arrive at the headphnes slightly late. That echo drives most people nuts. If yours doesn’t do that, then you win. But don’t go around telling everybody they can do it. That’s very unusual.
Most people have to live mix outside the computer to get perfect overdubbing sound.
I forgot to mention, the backing track OR one with a digital lead on it as well is already on disk and loaded into Audacity just as a guide/reference. This track may be just drums, or rhythm or a full digital cover, it doesn’t matter because it will not be used in the final mix (not as-is anyway). It does have a starting cue pip a few seconds past the track beginning. When I hit record to begin recording the (plugged in, not micced) guitar signal (no singing) this reference track starts to ‘roll’ the playback at the same time as the recording track begins to move. Sofar no problemo. As I hear the cue pip and the intro metro I get on step and jump in where the lead is supposed to begin.
I next try to play back the two tracks together, as I would to mix them although at this stage I don’t want to mix at all, I only want to chop the front of the recorded track to to start exactly at the reference track pip past the 4-5 second silence leading to it. There is however no point in trying to chop to the point where the cue pip is on the reference track because at this point thetwo are NOT in sync for some reason.
There are two problems, the first one needs fixing first.
1
The waves on the recorded track are leading the reference sounds, they are physically closer to the beginning
2
I cannot get a perfectly colocated cue pip onto the track being recorded
For the final mix all component tracks already exist on disk and begin exactly with the cue pip because I have chopped them beforehand so that just loading them assures sync. It’s the recorded lead track that presents both problems.
So what I do is first try to sync them up just on the basis of the wave profiles using trial and error, this is not easy when the lead part starts with poorly defined ‘pings’ and matybe not even near the beginning, I may have to look for alignment far into the piece. THEN I chop the recorded file as close as I can to the cue pip on the reference file and it goes into the bin containing all the other tracks waiting for final mix.
It’s a lot easier to use a longer click track lead-in instead of trying to jump right on one single pip.
Tick tick tick tick music.
The waves on the recorded track are leading the reference sounds, they are physically closer to the beginning
Did you set Recording Latency? That’s the one we can fix. This is a lot easier with a microphone, but tap your guitar with a pencil in time to the backing track. When you stop recording and look at the two tracks, they will be out of step almost certainly. Measure the miss and add that time to the latency value. Edit > Preferences > Recording > Latency. Try the sound test again.
Eventually, you will come away from a recording with all the tracks lined up (assuming there’s nothing wrong with your machine).
If you have enough hardware and cables, you can use a lead between Headphone Out and Stereo Line-In so you won’t have to tap the guitar.
The closest explanation I found is the one pasted-in below. It isn’t enough. I can understand how a playback would delay while the buffer fills up but I’m cueing myself on what I hear so that should neutralize the effect of any latency in playback. Moreover would not any latency affect my recorded stream the same as the playback one? If the playback latency is not the culprit then it would have to be the recording latency, and I could visualize that easy enough …except that it would lag and not lead.
I don’t understand how changing latency settings (there are 2 and not much explanation) could have a different effect on the playback and on the recording to result in an offset between the two. The settings are assumed stock and are “Audio To Buffer: 100ms, Latency Correction: -130ms”. Looks like I’d have to change one of these by 100ms.
The reference or cue file starts with a bit of silence, then 4 ticks (snare) which is OK. I can get on-step from there and the thing doesn’t even have to be in the computer. I can play it in the stereo while recording into Audacity. The purpose of the playback is just to hear a guide, a glorified metronome. For the sake of example lets say the 4 ticks are 1 second from one another and start 3 seconds into the track. These count-in as 1,2,3,4 and then the lead to record is another 2 seconds so starting on the 1st tick I count 1,2,3,4,5,6 and hit it on the 7th. Great, I carry on and THROUGHOUT the recording the two sound in perfect sync. It falls apart when I try to replay the two tracks together.
If I don’t use the computer for the reference track at all I still have the task of copying the the first tick onto it exactly at the right point so as to mark the track beginning. This is why I was hoping that having the ref track right above the recorded one would be some help …but for that it has to be in sync.
\
“Apart from leading to difficulties when monitoring, latency causes other problems too. Playback of existing recordings also suffers a similar lag, but in this case the audio recording software can do something about it. The software has to initially fill up its memory buffers before audio playback starts, so it simply waits until these are full before actually starting to scroll the screen display and starting audio replay. Although this means a short wait before anything seems to happen, it does ensure that the audio you hear is glitch-free and that the waveforms being displayed on screen are completely in sync with it.”
With Jack Audio, I can get the latency down to a hardly noticeable 23 ms on my budget laptop, though I usually use it with 46 ms latency. A well tuned higher spec machine should be able to manage about 11 ms.
Using the “latency correction” method previously descibed, even my old laptop will give sample accuracy.
I’m unsure what your aim is. If you just want to be able to effectively overdub in Audacity, the steps described in koz’s first reply (and in the manual: http://manual.audacityteam.org/o/man/latency_test.html) should be all that you need. If you want a more theoretical / technical understanding about the issues, then I can provide some additional information (subject to available time )
the primary purpose of a recording session like this is for me to produce let’s say a lead track with the starting pip in the correct place where all the tracks have them
Then I don’t see why this is an issue for you at all. The Audacity default settings should work perfectly well on Linux. The default sound system on Linux (PulseAudio) has greater latency than Audacity (and the latency in PulseAudio is not consistent), but is still perfectly adequate if you are not overdubbing.
I tried with jack, the only thing I could connect it to was ‘system’, and in Audacity I selected jack instead of alsa. It ended with my guitar being played on beat by me but sounding in Audacity about half a second late. It’s impossible to jam with a backing track this way so I gave up on it. The original condition as posted and involving a +/- 100ms lead in the guitar track being recorded is unresolved and remains unchanged.
What I have swung to instead is isolating the playback to a completely external device and either feeding that into the same headphone where I have to hear my own sounds or just letting it come through from ambient which is enough for the purpose.
This leaves the issue of post-recording syncing to other tracks. The present test exercise is 120bpm which I think gives 2 bars/second but I haven’t got to that part yet. The tracks are recorded with jack_capture while rosegarden plays and all include a cue pip at the start of the very first bar. That’s where they get precisely chopped to later in Audacity so all tracks sync. The lead is recorded in Audacity directly so since I ‘hit it’ on the third bar that note gets 1 second of silence in front of it to line up with all the others.
That’s as far as I got so far, I do have another (crash) issue in another topic.
Latency correction physically shifts the track after recording in the direction and by the amount set in the preference. Negative value shifts left, positive value shifts right.
If the problem is that the current -130 ms setting makes the recorded track 100 ms too far to left, have you tried setting Latency correction to -30 ms?
Didn’t know that, just found it in the manual under ‘your first recording’
And I had completely misread ‘overdub’ to mean that a new track is recorded including the monitored one as well as the one being played as opposed to only the played track being recorded as seems to be he case. What is meant I think is that the ultimate operation or final mix will be an overdub.
I most certainly will do that now that I understand what the latency setting actually does. -30 should be right on, but I’m still curious how it comes about i.e. I had a hard time with handling a note being recorded what seems to be before being played
I’ll be back later with this because if it works it’ll simplify my life a lot
For multi-track editors (such as Audacity), the idea of “overdubbing” is to create a new recorded track while listening to other tracks. Thus you can create multiple voice / instrument recordings, recording one track at a time with each voice / instrument on separate tracks. This then allows you to edit / process each voice/instrument independently. You can also use overdubbing for recording new tracks with a backing track. When the project is Exported to create a normal audio file, all tracks (provided that they are not muted) are mixed down to create a single mono or stereo file.