Making bit-exact WAV file copies of DATs

[PatMcGee]

Hi Koz,
OK, sorry, I wasn't reading right.

Yep, I certainly understand the environment that Audacity works in and the limitations that come from that.

Thanks to you and Bill, I now understand more about what's going on. I think my only lack of understanding now is why the mk II shows a certain error rate and the samples in the wav files show one or two orders of magnitude more in the number of differences. The only place left to consider is whether this might be something done in the OS X audio input driver. I don't know how to figure out if that might be the case, and if it is, what I might do about it. Maybe I can borrow a Windows machine with Toslink input and try that.

I certainly wouldn't go to this much trouble for any other (read: reasonable) project. But for the Library of Congress, I'm willing to do extra work.

Thanks,
Pat

[kozikowski]

I need to write a program that will compare files but ignore small differences.

Exactly. In the early days of Mac video editing, the ivory tower people designed Final Cut to insist on absolute perfect matchup between time code, frames, and audio sync. More times than I care to count, we had to resort to the far sloppier iMovie to get the work out the door at all.

I would probably only worry about the dropouts. The other errors are not significant. Nobody cares that the only pictures of the Hindenburg weren't perfectly exposed.

Koz

[PatMcGee]

Success at last!

I poked around in Audacity for other things I could change. In the Quality tab, I changed the Default Sample Format from 16-bit to 32-bit float. When I exported two different recordings of the same tape to 16-bit WAV, I got two bit-identical files (after I trimmed off the leader and trailer).

I'm not sure why that makes a difference. The input is 16-bit; the output is 16-bit. Why would saving that in a 16-bit sample cause rounding or conversion errors?

Koz and Bill, thanks much for your help. Without you guys, I'd still be floundering.

Pat