Full Version: 96 to 44.1 conversion problem
I have recently started to record at 96/24 with the FR2 Fostex field recorder. After having processed the file in Soundforge 9, I converted it on that platform to 44.1/16. Two separate steps are involved: Bit conversion and resampling. Both steps seemed to work well, but when playing back the 44.1/16 file, the pitch had been increased by about a semitone! I went through the process several times reversing the order of the steps but the result was always the same and was immediately discernible after the resampling stage. Any ideas???
You are probably playing it back at 48kHz by mistake. Check the sample rate of your soundcard and also check that the clock source is the soundcard's internal clock and not the clock from a digital input (eg SPDIF)
Also, you probably are doing it this way and wrote it backwards, but you should resample first and reduce bit-depth second.
Moderation: Removed useless quotation of previous post. What's the point when yours is the first reply???
Also, you probably are doing it this way and wrote it backwards, but you should resample first and reduce bit-depth second.
Moderation: Removed useless quotation of previous post. What's the point when yours is the first reply???
I friend of mine had this problem once before. Try to reinstall your soundcard drivers, avoid the Windows buildin drivers.
Very strange! I'd be really surprised if this was a bug in SoundForge, but it wouldn't hurt to try a different editor for resampling. (The math isn't that hard!)
The 96/24 file plays-back OK from SoundForge, using with the same soundcard/system as you're using for 44.1/16, right? All of the hardware is the same, and the only thing that's changed is the sample rate conversion?
In order to isolate the problem, you'll probably need to swap things around, one thing at a time... Try different software, try a different computer (or different soundcard), and try making a 24/96 recording on a different computer (to remove the stand-alone recorder from the process).
Usually, a pitch shift error is caused by "tolerance" in the soundcard's clock. But, that should show-up at any sample rate. So, usually this kind of problem shows-up when you record on one computer and play-back on another, or when you burn a CD made from a soundcard's with inaccurate frequency. If the Foxtex clock doesn't match the soundcard clock, I suspect the soundcard is the problem... But again, a clock mismatch should show-up before you resample.
You can also have a problem if the file header has the sample rate incorrectly labeled. But, that's very unusual... I suppose a flaky driver could write (or read) a file header incorrectly... Sometimes people intentionally alter the header to make the file play back faster or slower, but that's not the case here.
The 96/24 file plays-back OK from SoundForge, using with the same soundcard/system as you're using for 44.1/16, right? All of the hardware is the same, and the only thing that's changed is the sample rate conversion?
In order to isolate the problem, you'll probably need to swap things around, one thing at a time... Try different software, try a different computer (or different soundcard), and try making a 24/96 recording on a different computer (to remove the stand-alone recorder from the process).
Usually, a pitch shift error is caused by "tolerance" in the soundcard's clock. But, that should show-up at any sample rate. So, usually this kind of problem shows-up when you record on one computer and play-back on another, or when you burn a CD made from a soundcard's with inaccurate frequency. If the Foxtex clock doesn't match the soundcard clock, I suspect the soundcard is the problem... But again, a clock mismatch should show-up before you resample.
You can also have a problem if the file header has the sample rate incorrectly labeled. But, that's very unusual... I suppose a flaky driver could write (or read) a file header incorrectly... Sometimes people intentionally alter the header to make the file play back faster or slower, but that's not the case here.
QUOTE (DVDdoug @ Mar 31 2009, 23:03) 
Very strange! I'd be really surprised if this was a bug in SoundForge, but it wouldn't hurt to try a different editor for resampling. (The math isn't that hard!)
The 96/24 file plays-back OK from SoundForge, using with the same soundcard/system as you're using for 44.1/16, right? All of the hardware is the same, and the only thing that's changed is the sample rate conversion?
In order to isolate the problem, you'll probably need to swap things around, one thing at a time... Try different software, try a different computer (or different soundcard), and try making a 24/96 recording on a different computer (to remove the stand-alone recorder from the process).
Usually, a pitch shift error is caused by "tolerance" in the soundcard's clock. But, that should show-up at any sample rate. So, usually this kind of problem shows-up when you record on one computer and play-back on another, or when you burn a CD made from a soundcard's with inaccurate frequency. If the Foxtex clock doesn't match the soundcard clock, I suspect the soundcard is the problem... But again, a clock mismatch should show-up before you resample.
You can also have a problem if the file header has the sample rate incorrectly labeled. But, that's very unusual... I suppose a flaky driver could write (or read) a file header incorrectly... Sometimes people intentionally alter the header to make the file play back faster or slower, but that's not the case here.
The 96/24 file plays-back OK from SoundForge, using with the same soundcard/system as you're using for 44.1/16, right? All of the hardware is the same, and the only thing that's changed is the sample rate conversion?
In order to isolate the problem, you'll probably need to swap things around, one thing at a time... Try different software, try a different computer (or different soundcard), and try making a 24/96 recording on a different computer (to remove the stand-alone recorder from the process).
Usually, a pitch shift error is caused by "tolerance" in the soundcard's clock. But, that should show-up at any sample rate. So, usually this kind of problem shows-up when you record on one computer and play-back on another, or when you burn a CD made from a soundcard's with inaccurate frequency. If the Foxtex clock doesn't match the soundcard clock, I suspect the soundcard is the problem... But again, a clock mismatch should show-up before you resample.
You can also have a problem if the file header has the sample rate incorrectly labeled. But, that's very unusual... I suppose a flaky driver could write (or read) a file header incorrectly... Sometimes people intentionally alter the header to make the file play back faster or slower, but that's not the case here.
Thanks a million for all the suggestions. I will try them one-at-a-time!!
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.