So, I run on Windows with a Focusrite Scarlett 4i4 and this is generally how I set things up for vocal recording:
I have a separate Recording mixer track that takes the audio input and I generally only run a noise gate here. This track is then sent to a "Recording FX" track where I have a basic reverb (for mood
) and Pitcher or WavesTune which is in turn set to output directly to ASIO. The setup allow me to record completely dry audio if I arm the Recording track (with or without noise gate depending on if pre- or post-), or with FX if I instead arm the Recording FX track, with virtually no feedback delay (even with large buffers). Happy days.Now, obviously the recorded audio will be delayed and off grid as I have other playback tracks with generators and FX, nothing strange there, and as far as I can tell there are two ways of compensating for this:
1. Use the Input latency compensation to adjust the recorded audio (positive values, generally)
2. Set the track latency of the armed track i.e. "Set from...master"
Now, option 2 works pretty damn good actually as I don't have to mess around with entering values manually BUT it doesn't update if I modify anything else that affects latency. I have to "Set from...master" again. This is annoying but I can live with it and to be honest I don't think that this is the intended use, as it's more for compensating for ill-behaved plugs. But should it not, or could it not, be updated AUTOMATICALLY as I have said that I want to take it from master?
Option 1 is what I understand you SHOULD use to compensate for latency (event->all your plugs and routing->headphones->sing->sample) but how do you estimate the value and does it actually behave as expected?
So, I did an experiment, simply recording the metronome by sticking the headphones on the mic and recorded the result. I then opened the clip in Edison and measured the delay from grid to the start of the sample. I then entered this value into the "Input latency compensation" and recorded again. Now the audio is ahead of grid with more or less the same value. I divided the value with 2 and the audio is more or less on grid, voila! Now, there is always a 40-60 sample offset that I just can't seem to get rid of regardless which is also strange) but:
Is this the way it is supported to work, that I should divide by 2? And if so, why? I can mention that the calculated output + plugs delay in the audio settings is of no use here, from what I've been able to tell. Is there a better process to get this spot on?
Anyone?