Stuttering recorded segment

[cmac185]

In the last 6 months a very strange occurrence during recording has happened 3 times. A short segment of less than one second in length is repeated 3 or 4 times and the repeated segment gets longer and more clear with each repeat. I was there during the recording and the speaker did not do this. This was somehow generated during the recording. Whether it is something in Audacity or in OSx I have no idea, but am just posting this for information sake and to see if anyone else has ever seen this. It is a very rare occurrence and was going to just pass it off as "one of those things that can't be explained" until it happened the third time. It occurs somewhere in the middle of an otherwise normal recording and the rest of the recording is fine and other recordings done the same day are fine. Moon Phase issue????

The first two occurrences I believe were using a pre-release build of 2.1.2 and the last occurrence the GA version of 2.1.2 was being used. Computer 2012 MBP w/OSx 10.11.3. Audio input is wireless mic receiver direct analog to Line-in port on the computer.

Sound file attached.

[cyrano]

I've seen more than one case, with different audio interfaces, all under El Capitan. And as you say, it's rare.

I know it's a buffer thing, but have no clue what causes it. Since it can't be reproduced, I wouldn't know how to fix it. The Console shows nothing in the logs either. So I'm guessing it's a low-level OS X bug.

[cmac185]

Do you think any change in the Audacity recording buffer would affect it? I'm running the buffer at 15ms right now.

Cliff

[Gale Andrews]

Do you think any change in the Audacity recording buffer would affect it? I'm running the buffer at 15ms right now.

Probably not, if this is the bug Koz experienced once where fragments get duplicated and moved around.

It could be an OS X bug, but only Koz and you have reported it.

Note that Audacity has recently updated to the latest PortAudio release. So can you please try the latest Nightly Build which is http://www.audacity.homerow.net/index.p ... 0-919712e/.

I haven't yet got round to looking if the default latencies reported by PortAudio have now changed. If so, this would probably affect the required Audacity "Audio to buffer" setting, but I don't see anything directly relevant in the PortAudio commit logs.


Gale

[cmac185]

I'll give it a try.

Cliff

[kozikowski]

It's particularly entertaining if you take the problem upside down, my favorite diagnostic trick. If someone paid you to produce this effect, how would you do it? Rearranging segments of a voice recording in real time without apparently losing any content is very difficult to do. Yet somehow, the machine got it in it head to do it by itself, unbidden. In my case, the effect gets worse over time, so it's not even a static problem.

That machine has an air-gap firewall—no internet connection and it has an internal 112GB SSD with 9GB free. So virus, malware and speed issues need not apply.

Koz

[cmac185]

Interesting. In the 3 times I've seen it I've not noticed it getting worse unless you mean more often which probably is the case here as well which really is strange if that is the case.

My machine does have internet exposure, but I don't go to strange sites so wouldn't expect to have issues and viruses are virtually unknown as far as I know on OSx. There has been malware I understand, but I check for that occasionally using Malware bytes.

Just had a thought, I don't think it ever happened on my older MBP. This one is a 2012 with a I7 processor and lots of ram and 100GB free on SSD so as you said speed, memory etc. should be no problem. Could it be too fast? That would be wierd, but maybe with the I7 being effectively 8 cores running some of the multitasking between cores could mess up I suppose.

Ciff

[kozikowski]

I don't think it ever happened on my older MBP.

'frances' is a 2009 MBP.

I've never had it happen on the newer machines.

Koz

[cmac185]

Ok. My older one was a mid-2009 as well. Good machine. Went back and checked how long I've had this one and it's just over 3 months so it probably did happen once on the 2009. Guess we can toss out the speed idea.

Cliff

[kozikowski]

That's what I'm saying. It's almost impossible to design a way to force a machine to do this intentionally. So it's actually two mysteries.

Bug or Damage Resolution involves multiple people being able to replicate a problem exactly on multiple machines at any time.

This problem fails everything.

Developers have a condition called "Moonphase Errors."

"On a first Thursday of the month when the moon is quarter full and the humidity is above 73%...."

Koz