Jump to content

How much latency is too much?


Recommended Posts

  • Members

I have a Pentium 4, 3.2 running Sonar 4 Studio with an Audiofire8 interface. I've been running 24 bits with a buffer of 128. According to Sonar, this resulted in a 2.1 ms latency. This was fine, until I got above 6 audio tracks and I started noticing some clicks and dropouts in sound when recording.

 

So I changed the buffer to 256 on the Audiofire, which Sonar automatically adusted for, and now at the fastest Sonar buffer setting, it shows 5.9 ms latency.

 

So my question is this: Is 5.9 ms a big deal? I generally record only audio, overdubbing one track at a time. Does this mean that all subsequent tracks will be "behind" the original track (drum loop) by 5.9 ms? I understand latency represents about a foot per millisecond awy from the sound soure, but I'm wondering if this will matter when I go to mix 10 tracks or so to burn a CD.

 

Thanks.

Link to comment
Share on other sites

  • Members

Your tracks will line up with eachother. Your software will compensate for the latency. Latency generally becomes a problems when monitoring the recorded signal while recording. Vocalists are most sensitive to latency as they are hearing the sound within their own head, as well as the delayed signal through their headphones. Some vocalists are more sensitive than others.

Link to comment
Share on other sites

  • Members

 

Originally posted by raw-tracks

Your tracks will line up with eachother. Your software will compensate for the latency. Latency generally becomes a problems when monitoring the recorded signal while recording. Vocalists are most sensitive to latency as they are hearing the sound within their own head, as well as the delayed signal through their headphones. Some vocalists are more sensitive than others.

 

 

Thanks. But just so I understand fully: if I had say 100 ms latency, I would hear (and presumably be disturbed by) the lag time while overdubbing, but when I play it back during mixing the tracks would all line up? Why doesn't the software compensate during recording as well?

Link to comment
Share on other sites

  • Members

 

Originally posted by Bluewater



Thanks. But just so I understand fully: if I had say 100 ms latency, I would hear (and presumably be disturbed by) the lag time while overdubbing, but when I play it back during mixing the tracks would all line up? Why doesn't the software compensate during recording as well?

 

 

A "time machine" would be required. Maybe someday, but not today.

Link to comment
Share on other sites

  • Members

That's exactly why I don't recommend soft-monitoring. Go for zero-latency monitoring. What I mean is monitor your Input, not your Output. A lot of audio interfaces include this option as standard, but you can use a mixer if your soundcard doesn't have this.

 

All soundcards, converters, software have latency. That's unavoidable because it takes time to process the numbers. While it is possible to get that down to a few milliseconds, usually that is at the expense of system resources, and you risk clicky audio. I suggest you relax this a bit, and use zero-latency monitoring instead.

 

For this example, I'll assume latency is a huge 100ms to make my point.

 

How this works - typically you create your midi backing track, and off course when you play this back it will be delayed 100ms. This is noticed as a delay between pushing Play and hearing audio - but in most cases this won't worry you.

 

Sing/play your overdub and the audio file is recorded to your PC. Good recording software, like Cubase SX, will compensate for the delay, so this track is perfectly lined up with your backing track. The stored audio is time-stamped for synchronisation purposes, so this compensation works on subsequent playback.

 

However while you are actually recording and monitoring through the software (i.e. the soundcard Output) you will hear the vocal coming back 100ms later than you sing. That's unacceptable, even for Elvis, so don't do that.

 

Instead, take a spare output from your mic preamp, and using a mixer, combine the soundcard output (for the backing track) with the preamp output (for zero-latency vocal monitoring).

 

Subsequant overdubs are just the same - latency is of no real consequence.

 

Except when you are playing virtual instruments in real time, e.g. tracking. That's where you want to get the soundcard latency as low as is possible.

Link to comment
Share on other sites

  • Members

 

That's exactly why I don't recommend soft-monitoring. Go for zero-latency monitoring. What I mean is monitor your Input, not your Output. A lot of audio interfaces include this option as standard, but you can use a mixer if your soundcard doesn't have this.

 

 

I definitely recommend you follow this advice. I used to try and record guitar parts using soft-monitoring so that I could hear the effects of plugins as I recorded. However, at the time, a latency of 20 ms made it very difficult for me to give a good performance. A latency of 100 ms would be virtually impossible to get a good performance.

Link to comment
Share on other sites

  • Members

 

Originally posted by Kiwiburger

That's exactly why I don't recommend soft-monitoring. Go for zero-latency monitoring. What I mean is monitor your Input, not your Output. A lot of audio interfaces include this option as standard, but you can use a mixer if your soundcard doesn't have this.


All soundcards, converters, software have latency. That's unavoidable because it takes time to process the numbers. While it is possible to get that down to a few milliseconds, usually that is at the expense of system resources, and you risk clicky audio. I suggest you relax this a bit, and use zero-latency monitoring instead.


For this example, I'll assume latency is a huge 100ms to make my point.


How this works - typically you create your midi backing track, and off course when you play this back it will be delayed 100ms. This is noticed as a delay between pushing Play and hearing audio - but in most cases this won't worry you.


Sing/play your overdub and the audio file is recorded to your PC. Good recording software, like Cubase SX, will compensate for the delay, so this track is perfectly lined up with your backing track. The stored audio is time-stamped for synchronisation purposes, so this compensation works on subsequent playback.


However while you are actually recording and monitoring through the software (i.e. the soundcard Output) you will hear the vocal coming back 100ms later than you sing. That's unacceptable, even for Elvis, so don't do that.


Instead, take a spare output from your mic preamp, and using a mixer, combine the soundcard output (for the backing track) with the preamp output (for zero-latency vocal monitoring).


Subsequant overdubs are just the same - latency is of no real consequence.


Except when you are playing virtual instruments in real time, e.g. tracking. That's where you want to get the soundcard latency as low as is possible.

 

 

Thanks. It appears that the Audiofire8 does allow for hardware monitoring (the manual doesn't discuss this, though). I use the software Audiofire Console (mixer) to send my vocal input to the same outputs that the backing tracks from Sonar are going to. Presto. So then I understand Sonar will compensate for the 5.9ms latency during mixdown so it's not an issue there either.

Link to comment
Share on other sites

  • Members

woh, what do you have going on in that pic? i see a desk type thing with sticks, and a kick type pedal down below. i'm not too up on my midi technology, but is that some kind of whacked out psychadelic looking midi drum pad? just curious. nice fab four theme by the way. i also see the tape machine thats barely in the picture, definitely wish i had the money and knowledge to be into that kind of thing.

Link to comment
Share on other sites

  • Members

That tape machine has not been plugged in for 10+ years.

:rolleyes:

 

.the drum pad is from a TRAP KAT I play this contraption live and in the studio....connected to a Kurzweil rack....It's the best "invention" for lazy drummers and engineers!! We can track, in a small room > with no leakeage > and jjust edit sounds & performance, on the sequencer. Great live @ a gig to! Plug & Play...GROUP%20sm.jpg

Link to comment
Share on other sites

  • Members

 

Originally posted by Bluewater

I have a Pentium 4, 3.2 running Sonar 4 Studio with an Audiofire8 interface. I've been running 24 bits with a buffer of 128. According to Sonar, this resulted in a 2.1 ms latency. This was fine, until I got above 6 audio tracks and I started noticing some clicks and dropouts in sound when recording.


So I changed the buffer to 256 on the Audiofire, which Sonar automatically adusted for, and now at the fastest Sonar buffer setting, it shows 5.9 ms latency.


So my question is this: Is 5.9 ms a big deal? I generally record only audio, overdubbing one track at a time. Does this mean that all subsequent tracks will be "behind" the original track (drum loop) by 5.9 ms? I understand latency represents about a foot per millisecond awy from the sound soure, but I'm wondering if this will matter when I go to mix 10 tracks or so to burn a CD.


Thanks.

 

 

5.9 is not too bad. I run around 20 with no problems at all, untill it's time to track the MIDI drums. Then I bring things down to 5.8 and add 10-20 time+ (Cakewalk) on the MIDI drum track. I'm just getting started really but I've done a good bit with this level latency and it's really quite manageable.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...