otek
November 30th, 2006, 12:22 PM
I have just recently started dabbling with the Waves IDR plugin for dithering audio files. I am using Sony Sound Forge 8 for all my destructive editing needs.
This has brought up an issue which I'm kind of hazy about regarding the bit conversion process.
Previously, I have always used my DAW (whatever it happened to be) to make the conversion from 24 to 16 bits. Dither was always an option for this process, so it was my understanding that as long as a dither option was selected, the file would not be simply truncated (as in truncating the 8 least significant bits). I never bothered to properly investigate it, but always took this as a matter of course.
Now, when going about it the other way around, which is specifically applying dither to the signal without specifying bit conversion, the file tags in Sound Forge after this process still show the file as being 24 bit. The file size indeed also remains the same. When I save it, it gives me the option of selecting the original format, as well as various 16-bit alternatives.
So my question is, what happens to the file when I select to save it as a 16-bit file, after having dithered it? Will the dithering process work in such a way on the file that the 8 LSB's aren't compromised (or rather, less compromised) by the subsequent truncation?
I suppose the question can also be viewed as, what actually happens mathematically to the LSB's during a no-dither bit rate conversion? By applying a specific waveform (or shaped noise) to the audio waveform, is all the dither does to make the truncation of the LSB's less offensive, or does the actual bit rate conversion in and of itself involve something more?
Thanks,
otek
This has brought up an issue which I'm kind of hazy about regarding the bit conversion process.
Previously, I have always used my DAW (whatever it happened to be) to make the conversion from 24 to 16 bits. Dither was always an option for this process, so it was my understanding that as long as a dither option was selected, the file would not be simply truncated (as in truncating the 8 least significant bits). I never bothered to properly investigate it, but always took this as a matter of course.
Now, when going about it the other way around, which is specifically applying dither to the signal without specifying bit conversion, the file tags in Sound Forge after this process still show the file as being 24 bit. The file size indeed also remains the same. When I save it, it gives me the option of selecting the original format, as well as various 16-bit alternatives.
So my question is, what happens to the file when I select to save it as a 16-bit file, after having dithered it? Will the dithering process work in such a way on the file that the 8 LSB's aren't compromised (or rather, less compromised) by the subsequent truncation?
I suppose the question can also be viewed as, what actually happens mathematically to the LSB's during a no-dither bit rate conversion? By applying a specific waveform (or shaped noise) to the audio waveform, is all the dither does to make the truncation of the LSB's less offensive, or does the actual bit rate conversion in and of itself involve something more?
Thanks,
otek