I'm Dithering..
-
jasontorque
- Posts: 21
- Joined: Sun Apr 11, 2004 4:59 am
- Location: UK
- Contact:
I'm Dithering..
I usually mixdown at 24bit / 48kHz. I use an external mixer and bounce the stereo output back via ADAT, into Live as an AIFF. The thing that's confusing me a little is 'Dithering'. I know CD audio is 16bit / 44kHz, but I can still burn the 24bit / 48kHz stereo AIFF to CD without altering it, using Toast. Do I presume by this that Toast is doing the dithering, and if so, is it better to dither myself to a 16bit / 44kHz file before burning. I wondered what/when/where (or why?!) I am needing to dither at all..?
Urbantorque Recordings
RSS: feed://urbantorque.com/blog/backend/ut.rdf
RSS: feed://urbantorque.com/blog/backend/ut.rdf
I'll try and tackle this, bare with me.
Dithering relates to the sample rate. 24bit has a higher dynamic range (available difference beetween loud and soft parts) than 16bit. If you downsample to 16bit (what toast does when it burns your Cds) Material outside 16bit dynamic range are to0 quite sounding and finely detailed to be represented by 16 bits of info so they're lost. Dithering can save some of this info by adding randomised noise into the mix at a very quite level (you might see an option for 1 or 2 bits of depth). This noise combines with the quiet information making the total mix slightly louder at those very quiet points, just loud enough that it can be represented by the 16bit sampling rate.
The upshot of this is you percieve more dynamic range than you actually have. Where you have a range of 0db to -96db, with correct dithering it will sound something like 0db to -105db. Dithering Noise cant really be heard when its mixed withanother signal because it sums to zero mathematically.
This is my memory of Bob Katz's explanation in his book on Mastering (Beg, borrow or Steal a Copy). His personnal recomendation is triangular dithering
Dithering relates to the sample rate. 24bit has a higher dynamic range (available difference beetween loud and soft parts) than 16bit. If you downsample to 16bit (what toast does when it burns your Cds) Material outside 16bit dynamic range are to0 quite sounding and finely detailed to be represented by 16 bits of info so they're lost. Dithering can save some of this info by adding randomised noise into the mix at a very quite level (you might see an option for 1 or 2 bits of depth). This noise combines with the quiet information making the total mix slightly louder at those very quiet points, just loud enough that it can be represented by the 16bit sampling rate.
The upshot of this is you percieve more dynamic range than you actually have. Where you have a range of 0db to -96db, with correct dithering it will sound something like 0db to -105db. Dithering Noise cant really be heard when its mixed withanother signal because it sums to zero mathematically.
This is my memory of Bob Katz's explanation in his book on Mastering (Beg, borrow or Steal a Copy). His personnal recomendation is triangular dithering
-
jasontorque
- Posts: 21
- Joined: Sun Apr 11, 2004 4:59 am
- Location: UK
- Contact:
..
I would use an editor for dithering becaus with Live dithering will take place before master level and that's not what you want.
Btw some people say that with 32 bits you should always dither when exporting to a fixed bitrate, not only 16 bit, but also when using 24. Floating point causes truncated errors which might cause peaks in the lowest parts of the signal. I have to say I never had a problem with it myself.
I dither in Wavelab4 btw. Great editor.
Btw some people say that with 32 bits you should always dither when exporting to a fixed bitrate, not only 16 bit, but also when using 24. Floating point causes truncated errors which might cause peaks in the lowest parts of the signal. I have to say I never had a problem with it myself.
I dither in Wavelab4 btw. Great editor.