Ableton Forum

ok, im running a fast computer, g5, 2ghz, plenty O ram, tropical fish mouse pad..

but im using an M-Box,

When i record from the ext in, everything is shifted a couple mileseconds off the beat. I always used to think it was my playing, but im sure its the latency. i have the settings in the audio preferences set to

output latency 11ms
input latency 11mis

Settings
overall latency 11ms + 11ms +-22 ms = 0 ms


how can i get this fixed?

when i record, i mute the track, and set the mix on the mbox somewhere between playback and monitor. im recording from turntables and bass guitar.

Thanks,
Bernie

P.S. whoever fixes this problem gets to stay at my wonderful beach house for the weekend.

are you recording in 256 or 512 in ur buffer settings?
22ms is the problem with ur recording being out of time..you should update the drivers if you havent, and bring down the buffer size..

The way I do it is to record dry in live, with monitoring off (on the performance track view). Monitor yourself through the Mbox. Latency compensation only works this way.

Also, in the audio settings, make it add 11ms, totaling 33ms.

I tested my method with an external midi synthesizer, looked at the total delay, and added that time to the adjustment setting.

There's a routine described in the Live manual. In short : send a quantised audio track like a 4 on the floor kick or short snare to the outputs of your mbox, which you connect back to your inputs. Arm record a track with your inputs as audio ins. Record a clip, then zoom on it to measure the average gap between the grid and your newly recorded audio. Add this to your latency compensation value.

The way I do it is to record dry in live, with monitoring off (on the
performance track view). Monitor yourself through the Mbox. Latency
compensation only works this way.

Also, in the audio settings, make it add 11ms, totaling 33ms.

So this latency compensation param actually works in this way?

From the way it shows the equation I always figured it was for adding
EXTRA latency so as to sync with an even slower piece of gear (i.e.
another laptop) thus making more latency, but in sync with the slow
gear...

How does it work?

Does it really move the start of the clip back after recording so that there
is zero latency...... Wasn't there a 5 page thread with people asking for
automatic compensation.... I can't believe this feature actually already
exists if people demanded it so strongly....

I honestly believe it just adds MORE delay to the overall mix rather than
moving new stuff back to be in sync.

It's been awhile since I did kindie math.... but 33ms don't = 0ms in my book.

-Ben

no ben you had it right the first time.
it adds latency to your playback material so that your playback material matches up with your input/thru material.

Yup... ok....

So in reality it's not really useful for syncing live (played instrument) input.

Using 11ms as an example....

If your playback is 22ms delayed
Then your input is 11ms delayed

Then you're just getting further and further away from 0ms.

If you play along with what's coming out in the speakers then there is no
way you're magically going to be in sync unless you compensate in your
head/fingers.

Push the playback back 11ms and play your instrument along with it and
you'll just be another 11ms late.

are you recording in 256 or 512 in ur buffer settings?
22ms is the problem with ur recording being out of time..you should
update the drivers if you havent, and bring down the buffer size..

So for the benefit of not confusing Bernie

This is the root of the problem, sample buffers are the only way to
reduce latency. The smaller they are the more chance of clips and pops.

Also you'll probably get used to playing ever so slightly ahead of time....

If you externally bounce something around a few times (I use Live3) then
it will get really bad so you go into the clip view and manually drag the 1
back to where it should be. However if using this method be aware that
you'll need more space on the end of the clip if you want to keep it the
same length.

i.e. If you have a 16 bar loop and you move 1 over a couple of ms
chances are you won't have enough room to squeeze the last note in. You
can warp stuff a bit to fit.... or you can just record 16+ bars instead then
you'll have room to move.

-Ben

the problem isnt with the MBOX it's with Live. Unlike just about every other audio software out there, live does not trauncate the beginning of the clip to remove the equivelant of whatever you buffer setting is.

The reason direct monitoring was invented is so that you could have high latency, but as long as it sounded right when you direct monitored, the software would make it sound the same as that.

I don't often criticise ableton very strongly, but this is a no brainer which makes me go WTF were they thinking?

Same....

Simple feature with massive appeal.

This coming from someone at 3 to 6ms depending on instruments running....

I'd die at 11ms!

-Ben

you could record at a higher rate - 96 - this will half your latency

Simple feature with massive appeal.

I'll revise that.... "simple".... I know how frusting it can be when someone
who doesn't know the code says something is "simple"....

Chances are it's actually hard because of the "not having room at end of
loop" problem I described..... Maybe the Live engine doesn't have a cache
of incoming signal past where it stops recording, which would make
this "simple" feature actually mean a total redesign of some fairly deeply
rooted code.... or at least adding a tail of cache (that isn't displayed in the
interface) onto every recording so it has room to work with.

-Ben

MrYellow wrote:

Simple feature with massive appeal.

I'll revise that.... "simple".... I know how frusting it can be when someone
who doesn't know the code says something is "simple"....

Chances are it's actually hard because of the "not having room at end of
loop" problem I described..... Maybe the Live engine doesn't have a cache
of incoming signal past where it stops recording, which would make
this "simple" feature actually mean a total redesign of some fairly deeply
rooted code.... or at least adding a tail of cache (that isn't displayed in the
interface) onto every recording so it has room to work with.

-Ben

With all due respect this actually is quite simple. Since you know the buffer size when you begin recording, you can put that much digital silence at the end of a clip. In addition to that they could just modify it so the "offset" value does the trick, or the first warp marker goes in. They have so many options.

I don't wanna sound all "gimme" "gimme" but this is not a "feature" this is a neccessity. If you're going to have a "record" button, this is one of the things you need to implement in 2005. I don't know the facts, but I'm assuming programs like Cubase and Pro Tools have had such functionality for well over 7-8 years.

sweetjesus wrote:
I don't wanna sound all "gimme" "gimme" but this is not a "feature" this is a neccessity. If you're going to have a "record" button, this is one of the things you need to implement in 2005. I don't know the facts, but I'm assuming programs like Cubase and Pro Tools have had such functionality for well over 7-8 years.

TRUE !!

sweetjesus wrote: The reason direct monitoring was invented is so that you could have high latency, but as long as it sounded right when you direct monitored, the software would make it sound the same as that.

......so the monitoring button taces care of the sample-place adjustment ??



then how about the difference between Monitoring Auto / Monitoring On / Monitoring Off ....

I am lost here, examples please

I know nothing about MBox... but think he's talking about monitoring
directly from your sound card rather than the output from your
software.... Like with the RME I can 0ms monitor and make sub-mixes.....
However all that is meaningless when what's being recorded isn't exactly
the same as what you're hearing. Also monitoring from your soundcard is
kinda pointless when you're using VST effects on a live sound. No use
monitoring a dry bass sound when what you're recording has reverb,
compression and distortion...Sure no matter what you do there will
always be latency in a live sound. It's what the software does with the
sample once recorded that matters.....

With automatic compensation you can monitor (via Live) as you record
then only send to FOH once it's adjusted and looping..... It's impossible to
compensate in real-time without a timemachine...... or very good
headphones that block out all external noise and are ahead of the main
mix.

One thing I'd add to any requests of automatic compensation as a feature
tho is that it be switchable and customisable per track/input/channel or
whatever. This way when you're using heavy effects or something that
adds lots of latency, and compensating with your brain... you won't end
up with Live moving it on you. As you'd be able to turn it off for that
track, or maybe even set a different time......

um.... sorry it's getting late just adding a bit....

I can't complain too much, 3 to 6ms is nothing really..... but using Live3
and doing external bounces makes it frustrating when really the latency
should be compensated automatically. For really tight bass lines that have
to be on 1 *exactly* it's a pain too....... Guess the main thing is,
sometimes everything being exactly on 1 is very important.... If you're a
DJ style player then all you do is spend hours tweaking your tracks.
However if like me you want to do a lot of stuff on-the-fly.... Latency
makes it just that bit more difficult to get a tight mix.

When I'm playing beat 1 at 2.578 it don't matter too much tho

-Ben