live is not sample accurate!

[slow riot]

and yes. this probably isn't related in a major way to the topic at hand, which seems to be more of a glitch thing than a full major functional decision that would majorly affect behaviour like the issue I raised is.

sorry for the hijack.

[starving student]

Everyone not interested in the thread topic is dismissed from class today and will enjoy an extended recess, but everyone else who is interested in the topic of Live and sample accuracy stay right where you are, don't even move a muscle or suffer the consequences No not my consequences my friend but you know who's..............

[mr.ergonomics]

First I have to say I found out something new. I don't have the time to do a systematic test with all different options in mind, but I played around a lot with different options (route track 1 to track 2; record the audio in track 2 with "resample", ...).

When I record without "punch in" the problem don't appear in arrange view.


assumption : when you record something with "punch in" in the arrange view, your recording is 1 sample delayed.

Take a look....

needs -1 sample delay in track2...



perfect sample accurate...

so which warp mode are you using.

repitch

I just did it 10 times with monitoring on (on one recording channel) and monitoring off on one recording channel and was able to cancel out via phase reverse every time. That is scary if it is happening sometimes though.

Can you describe it a bit more detailed? I'm really able to reproduce this behaviour with or without monitoring on. You had "punch in" off, right?

I have a friend who is particularly sensitive to rhythm who says that live is not especially tight (I think he said clips mode was a bigger offender).

i have noticed it several times but do not trust myself enough to rule out the placebo effect and have not verified this information with recording/cancellations etc etc.

maybe, maybe not. senses can be a good indicator for something, but I only believe something if I have a prove. by the way you could easily do a "tight" test. just record it and compare.

[Fizmarble]

I just wanted to say I love the scientific approach you have mr. ergonomics. I am locked in to this subject til it is sorted.

[slow riot]

just record it and compare.

I think that is probably the best way to find out. thanks for your advice.

[mr.ergonomics]

Update 2:

Unfortunately the reason can't be the "punch in" thing alone. I had an idea and did the test again.
This time I didn't have the problem even with "punch in" on. I was a bit puzzled because of this so I did the test again a couple of times.

Result: Sometimes this problem occurs, sometimes it does not, it's seemingly sporadic.

Really just guessing, but maybe there is a certain sequence of functions in the code (...which isn't triggered everytime , therefore seemingly sporadic), which forgot to pass a value or something. I think I will stop testing at this point, because I really can't do more than reporting this strange behaviour. I hope Ableton take a look at this problem.

Again, I don't want to mock Ableton (actually I like them... ), but maybe this problem is not limited to this recording thing. Keep in mind, there is a similar problem in session view too.

[comaduster]

Strange. I still can't reproduce it the way you are doing it.
Before I wasn't recording with punch in or punch out on.
But I tried re-recording to another track, as you did in your description.

I WAS able to create non-canceling waveforms by doing the exact same thing mr.ergonomics, but rather than just recording a one shot I let play back run and intermittently turned global record on and off (to create segments of re-recorded material). Some of the segments would cancel, and some wouldn't. Very strange indeed.

I'll get some pictures posted ASAP.

[comaduster]

That was with punch in on.

[comaduster]

Doesn't seem to matter though.

[Hermanus]

Are you REALLY sure the problem is not linked with your audio card settings...
When I was recording with directX/MMe audio driver I've always encountered shifted recordings.


Since I've switched to ASiO, no shifts anymore, straight recordings.

Just my 2cents.

[mr.ergonomics]

no, unfortunately not. I use a RME multiface (ASIO of cause) and tested it on a computer from a friend too. I don't have problems like this in other software and I have some confirmations on this problem by other users too.

[zigzag]

i think its due to the authorization code kicking at odd times

[Parametex]

Ok heres my test ...

I recorded the sample as above and started the cancellation test and the results depend where I start the playback from!!!

If the audio playback is started from the exact beginning of the two samples they DO NOT cancel out. If the playback is started from a bit earlier they DO cancel out fine.


This bug (I trust its not a feature ) is very much in the root of the audio engine we all love and use. Such fundamental flaws should not exist if trust is to be obtained to from the users.

Thanks for the test to mr.ergonomics!

[zigzag]

has anyone contacted support about this? will you do it with this new evidence at hand?

[Parametex]

Maybe Mr Ergo could post the "official" bug report?