I am compiling information on hardware combination latency. Here is what I want to measure: feed your analog output from your DAW (where you monitor your recordings) directly into your mixer input. Playback a spike (or other easily identifiable waveform) from the workstation while recording this signal on a second channel. Now count how many samples the 2^nd signal lags behind the first. From that and the sample rate you can calculate the amount of delay.

Is this important to you and your recordings? Maybe, maybe not.

• If you record a project in one take – no problem – all tracks are delayed by the same amount.
• If your total latency is plus or minus (yes, your audio card’s driver can make this negative) 1 or 2 ms and you do a couple of overdubs – you’ll probably find your recording tight enough.
• If you do several multi-track layers and you’ve got a few ms difference or more between successive tracks – you could be getting screwed. Track by track it sabotages your work – here is what happens and it is insidious.

Lets say you work by yourself laying down one track at a time. You record a first track it’s a keeper. Second track is off by 3 ms – sounds good… After laying down several tracks you may find things getting sloppy What is happening is you are getting your rhythmic cues from different tracks. The 3 ms delay when the second track is recorded may be imperceptible (and at this point it may be desirable – after all, bands create music in real space – where sound propagates about a foot per ms) but by the time you’ve recorded the eleventh track the slop between rhythmic cues can have grown to an obvious 30 ms. I was burned by this a few years ago and eventually had to drag each new track ~286 samples to the left. What a pain.

The same thing can happen in the analog world just by being a several feet from your monitors (having a negative latency could work in you favor under this circumstance). I seem to recall Bruce Springstein dumping the e-street band and putting out a solo album that sucked. Could this partly be responsible?

For those of us who work track by track; monitoring with headphones and zero latency while recording is probably the best means of keeping your mixes tight.

Given the possibilities of hardware combinations and flexibility in signal routing within digital mixers, a wide variety of error is possible.

What can be done? The only solution that seems practical to me is an alignment parameter in the audio card’s device driver that would throw away a certain number of extra samples when a start recording instruction is received from your DAW software (if you know what you are doing you are probably monitoring the track being recorded in the analog realm anyway so who cares if that info gets to the DAW a little late). This could minimize the problem in all of your programs with one adjustment (needed when you change your hardware configuration). It is possible that some drivers may already have this ability - at least internally - but in this scenario it would have to be user adjustable.

Is there a problem? Well, that is where you come in. Collect the data. We need to know:

1. The amount of delay while recording a monitored playback track (in ms ).
2. The sample rate.
3. Your card make, driver & model.
4. Digital mixer make & model (or external converter).
5. DAW software.

Note) The problem can also exist in a conventional audio card but, by now, most should be pretty tight. For instance, the AdB card that I’m currently using has a latency of about –1/2 ms which I have found to be acceptable (2^nd track precedes the original by 28 samples at 44.1 kHz).

John Harragin