Ableton Forum

hi there,

i'm new here, and i thought maybe you knowledgeable people can help me figure something out. i tried something really simple: sessionview, playing a 1 bar midi-loop at 120bpm, four quarter notes, triggering a click sound in the virus a.

when turing on Live's metronome it is very noticeable that the virus click sound is constantly slightly delayed. Of course, its my soundcard's latency which is set to 4+4=8. but i would have expected 8msec to be much less noticable...

so i recorded these clicks as audio into session clips to check how high the "real world" latency of the soundcard-input is, plus the time it takes for the midi to travel and the virus to do its thing. to my surprise i end up with a 22msec offset (being late) on the quarternote-markers in the recorded clip. i recorded 4 bars and checked at several points in time to make sure it wasnt a "first its off, then it stabilizes"-kindof issue.

of these 22 msecs, 4 msecs are explainable, theyre the stated input latency of my soundcard in Live. the remaining 18 ... well, here's my questions:

1) is it possible that the actual latency of the interface is like 4 times higher? i didnt do the loopback test yet, but i will. or is midi transmission *that* slow? i cant imgine this.

2) how can i compensate for this? i know, there's several threads on latency/sync issues, but i didnt find answers and in general there seemed to be confusion when it comes to this. to be precise: this is not about latency when monitoring the virus (or any external synth) through my soundcard, and how to deal with this.

i would just like to find a way to tell live either "Play any MIDI 22ms early" (i know, the first note would be cut off etc) or rather "Move any recording back in time 22msec when done".

From another thread i gathered the latter cuold be done by adjusting the "Overall Latency", but no matter what i enter, even +50 or -200, in the recorded clip the audio stays 22ms late. the other related thing, "Midi Sync" is no use since im not talking about syncing an external sequencer or arp to live. still, i tried it, selected the virus port, entered 22, 50, -100, but again, no difference.

3) a little q in the end which i couldnt figure out, how do i disable snapping to grids for the start-marker in a clip in session view? i thought i just move the start marker ahead 22 msec, but it insisted on snapping to grid.

thank you for your patience,

aex

I asked a simliar question in another thread. The general consensus was that the problem lies in my audio interface, not with Live. Since I sometimes use audio loops (moving away from that for a hardware sampler) is to add a simply delay 100% wet on the audio tracks. Then they sync up with the MIDI devices. Live's metronome is still off, however. It's not a problem if you're just recording synth loops, because you can always slide a warp marker over a little bit to the right after you're done recording it.

There doesn't seem to be a way in Live to send the MIDI out earlier than anything else. I don't know if any other programs even do that. Every program has its limitations though, and I'd say if not for this, Live would be perfect. It works great for me after employing a few workarounds like the simple delay on the audio tracks, or just training myself to play the synths a little earlier!

Another solution would be to get a hardware mixer and not route any audio from your synths through your audio interface. That would solve the problem since it's your interface introducing the lag. However, if you wanted to record or apply effects to anything your routing might get a little creative with sends/returns on your hardware.

I've actually got a live gig coming up where we will be relying on MIDI loops, and we're going to bus everything through an external hardware mixer. The only things going into the audio interface will be a microphone for effects, and perhaps a guitar with some delay on it. I'm not expecting any sync issues with that.

Thanks for the quick response.

So if the general consensus is that the audiointerface must be the cause, its in this case the tascam us122 causing 22ms input latency instead of the 4ms it claims to have in the Live Prefs. this would be a poor performance indeed. i really have to do the loopback thing to check if the same delaytime comes up ... well actually it would have to be even larger than 22msec, taking into account the output latency as well. hmmm.

the workaround with using delay plugins on audiotracks, like setting them to 22msecs will indeed allow me to *monitor* external stuff in sync with the audiotracks, but once the external stuff is being recorded as audio, the newly recorded clips will have the 22msec delay "built-in", so for those i must make sure not to use a delay plug, i fear this can get messy. also, if i ever decide to change the latency setting in Live's prefs, and open an old project with lots of delay plugs tuned to my old latency value im gonna have a problem. in general, i'd rather sacrifce timing while monitoring the synths, and have them recorded accurately instead.

i'm not sure if the hardware mixer thing would help me other than with latency free *monitoring*. in fact i do have a little hardware mixer, all my synths pass through it and if i patched the mixer's monitor outs to my speakers, and feed Live's audio output to the mixer, i'd hear the synths next to live's audio in correct timing. but this way i cant record the synths into live at all, which rules this option out. would be a different story if i had more than 2in 2out on my tascam i suppose.

ah, and you mentioned the 1st warpmarker which i could slide to the beginng of the recording, hmmmyeah, it would kindof work, but then the clips's overall timing is changed then, isnt it? i figured moving the startmarker would be more elegeant, since nothing gets warped. if i record a 1 bar loop, i would of course have to record two bars of it and only use the second, to provide for seamless looping. you dont happen to know a way to make the startmarker move freely, non-quantized?

thanks for you time.

update: i did the loopback thing, connected the ins and outs of my tascam, turned monitoring off in the recording track in live, started a 1 bar loop on a different track, recorded one bar of audio, zoomed into the two loops and compared when the same transients occur in both of them. in the recorded loop everything was 10-11 msecs late, which is cool coz its close enough to the 4+4=8 setting in Lives Prefs.

so the audiointerface is not the problem.

i should be ending up with 5msec input latency plus a little for midi trransmission, instead i get 22msec. i would really like to know why this is, does anybody have clue?

i have the latest drivers, the midi-interface is a motu micro-lite which is not a cheapo one, i tirggered both the virus and the modular with the same result.

is this a live issue? or an osx issue? i dont own other hosts, i guess i'll have to grab some demos and see how logic etc. behaves in this respect.

I solved this problem by turning off monitoring in Live and using ASIO direct monitoring in my interface. Everything stays in sync now.

hey kuniklo, youre my hero

i have no idea why turning monitoring off on the rec track helps, but it does! and with this setting, the "overall latency" thing works too, adjusted it to 22 msec and now the clicks are nailed to the millisecond.

i had never used the direct monitoring feature of my interface, i only realize now how handy this is.

thanks again, you made my day - or night, over here

I'm going to try that tonight too.
Hope it works! Fun times await.

hi the_planet,

good luck with it, pretty sure it will do it for you.

i'd be interested to know how many millisecs of OverallLatency you end up with, and how this compares to the amount that your audio interface is - supposedly - set to ... i have 4+4=8 as the theoretical value, and 4+4+14=22 as a realworld value. yikes!

the audio-loopback test i did before had yielded like 11ms from out to in, BUT i only realized later that of course i had monitoring in live set to OFF - otherwise i would've gotten a feedback loop. so therefore the 11ms result is AFTER live had already compensated according to its OverallLatency setting, which was at its default setting at the time, meaning identical to the 4+4=8 i had entered for the "theoretical latency".

i was thinking a lot about the way live handles latency, like why is the OverallLatency thing only supposed to work when you dont monitor through live? the more i think of it it seems to make sense, but i'd be happy to hear if someone has seen right through it all. i cant make much out of the info in the manual when it comes to this.

ax