Sync Source and Destination ?

[Craig Maier]

Is the sync issue sensitive to things like sample rate and/or bit depth or a combination of the two? I ask this because there are buffers involved with each filter and one could be overflowing under certain file conditions.

Craig

[Craig Maier]

When Rick and I were on our daily business phone call, I asked him about this. He asked if the Preview Buffers has any impact on this. The Preview Buffers are located at Edit/Preferences/Soundcard/Preview Buffers. Rick also expressed that CNF fft size could have an impact on something like this for a given sample rate/bit file.

Craig

[Dan McDonald]

I was able to get the time offset by increasing the buffers to 50 (the maximum) on this computer. At the default (7), I can't get the error reproduced. Anyway, that should help track it down.

Dan

[Craig Maier]

Hi Dan,

Thanks! Interesting. It's counter intuitive, but it is a strong hint about the bug.

Craig

[Marc Hildebrant]

I use the default, 7, buffers for the sound card.

Marc

[Craig Maier]

Does reducing the number of buffers have any effect on the delay?

Craig

[Marc Hildebrant]

Craig,

More or less seems to have no effect. These are the buffers in the sound card tab.

Marc

[Craig Maier]

Hi,

Rick thinks the preview buffer change must be a 'side-effect' of the bug. It is not an easy one to track down. It is not consistent system to system. Not sure what to do.

Craig

[Dan McDonald]

This should help (although maybe it just makes it more puzzling).
I was able to reproduce the sync issue at my other computer, using 7 buffers.

I took the multifilter preset and ran just the file change, so I could take that out of the equation. No sync change with the file change filter.
Then I pulled the filters out, starting at the back, Still get a sync change without cnf #3
I pulled cnf#2 out, and the sync change went away.
I added cnf#2 back in, and got the sync change back.
I pulled cnf#1, and the sync change went away.
I added cnf #1 and #3, but took out #2, and the sync change was back.
I pulled all the filters out except the cnfs, and re-established the sync problem without anything else in the multifilter.
I get the same pattern, so that whenever cnf #1 and any of the other 2 are in the multifilter, it adds length to the file; it adds a constant amount, whether it is 1+2, 1+3 or 1+2+3 (in this case, on a file that is 02:47:0900 it makes it 2:47.1582.
I then opened cnf #1 and changed the fft size to 2028 (which is the same as the other cnfs), and the problem went away. No other changes were made.
I then changed fft size back to 8192, and the problem came back.

So, I think, the 8192 fft size does not play well with the 2048 fft sizes in the other cnfs.

whew!

I hope that helps.

Dan

[Craig Maier]

Whew, indeed!!! Thanks Dan. It sure is a complicated bug, but the fft size does play somewhat logically into the bug. It seems like it is an interaction issue with other filters and the CNF when the CNF is set for high ffts. We will have to look at that situation.

Craig

[Marc Hildebrant]

Thanks Dan for trying the different experiments with a fresh set of eyes. The 8192 number was a recent change that I added to the first filter and didn't "click" in my mind as a potential as much as it should have.

I did try 2048 in all three CNF blocks and on my system, I still have a time offset.

Marc

[Craig Maier]

It sounds like an extremely complicated bug. Not sure how easily it is fixable, if at all.

Craig

[Dan McDonald]

Very complicated. I'm surprised you got the bug with all three CNF blocks at 2048. I didn't get that. I'll see if it happens this evening, for one last test.

[Dan McDonald]

I just checked again, and I don't get the problem if I have all three set at 2048.