'Reverse latency' - is there such a thing? One Audio Masterclass visitor seems to have a problem...
Among all the uncertainties of computer recording, one thing is certain indeed - each reconfiguration of hardware or software will bring a new problem you have never encountered before.
And so it is with 'reverse latency'. Here is the message received by Audio Masterclass. The sender's first language is not English, by the way.
If anyone can shed any light on this, we will be happy to publish all constructive replies.
Here we go...
"First I program a drum set, usually with Fruity Loops, which is so far all no problem. Then I use a small neat software called "Internet Audio Mix" for multitrack recordings of guitar, bass and vocals.
It does not have much options, all it can actually do is mixing multiple tracks (I never found out so far where its limits are, probably it can do a hundred tracks or so). It also can envelope volume of the tracks, or pitch shift. That's all and so far I have always been happy with the simple and easy way to use it.
In the options it allows to select separately buffer sizes for recording and playback. Interestingly enough, these had so far totally no effect on the latency (neither had the settings in the asio4all tool).
Okay, so I launch Imix, load the drum set into it, plug the guitar into the mixer and turn on the microphone channel of the soundcard (line in and mic are weirdly merged in this one - but it has 3 line outs which I 'd never have a use for, since from the line out it just goes to stereo set).
So then, I put on headphones for monitoring from my stereo set. I start then the recording in Imix, and I can select the sound quality there, usually I use 44,100 hz, 1- bit. Finally I click the nice red button to start the recording.
Then I begin playing, and what I hear in my headphones is perfectly synchronous with the drum set which runs under it.
I must add, most tracks I make, are quite lengthy. So after about 10 minutes or sometimes 30 minutes, the drum set ends. I then let guitar ring out from the last drum hit, and stop the recording, entering a filename, a save-to location and then the thing saves, being afterwards added to the mix.
Now when playing back the entire thing, I notice a significant delay summing up over the course of the track more and more - but its the mentioned 'reverse latency'. That means where I strummed the guitar the last time to let it ring out, the drums are still going on, even though they should end exactly with the guitar.
That has happened about two weeks ago first time, on a freshly set-up computer. It has a 4 GHz processor, 1 gig of ram and well - the achilles heel seems to be the soundcard, a Soundblaster Audigy 2 SE.
Though now, the problem became even weirder: I tried then last weekend to put in my old soundcard again, a soundblaster 128 PCI , serving me well throughout almost 8 years I guess, which - surprise, surprise - did not deliver any sound output at all.
I reinstalled the new one, and I was surprised once again: the metronome test produced a perfectly aligned sound! Each metronome click was perfectly recorded where the programmed one was.
Full of joy, having thought the problem may have been solved with reinstalling the new soundcard and its latest driver update, I programmed some drums, loaded the result into Imix... and voila - I got the same latency again.
The problem just anyway is not a problem to exist in Imix alone. Also in Adobe Audition, I have exactly the same latency - recorded files 3 bars too early stopping.
Now to sum it up: If I am recording things, the monitoring I hear shows 100% synchronous playing (well, make that 95%, I am not such a good guitarist). Anyhow, after having saved the recorded file, it seems as if I had played too fast, recorded tracks ending way before the programmed drum set ends.
Now I must excuse if this all sounds still quite unclear; English is not my first language, and I probably know a lot of the problems and their technical terms just in German, if at all. Anyway I hope this helped to describe the problem more precisely."