Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[William Boyle AKA Elfy]

High party people. If i receive a stereo wave in 32 float. When importing to a 24 bit environment, should i dither.
Is the difference too small to debate.

[lucidwaves]

If I remember correctly (someone please correct me if I'm wrong) 32bit float is about the same in dynamic resolution as 24bit fixed and there would be no need to dither.

[Ronny]

lucidwaves wrote on Fri, 30 September 2005 10:03

If I remember correctly (someone please correct me if I'm wrong) 32bit float is about the same in dynamic resolution as 24bit fixed and there would be no need to dither.

32 bit has -48dB lower noise floor or more dynamic range than 24 bit, but that's theoretical until you are in the digi realm and processing from digital black. You are never really lower than the noise floor of the microphone on non black tracks and the best Neumann mics only have -81dB noise floor.

All you need to remember is to dither all wordlength reductions. This means when working at 32 float and returning to native 24 bit, you should dither. Now whether you can hear artifacts from a 32 to 24 bit truncation, is debateable, but there is always going to be less damage to a WLR when truncating without dither than with dither.

[William Boyle AKA Elfy]

Thanks for the help.

Ronny wrote:

but there is always going to be less damage to a WLR when truncating without dither than with dither.

Would you dither 32 float to 24 bitRonny?
Thanks

[Jack Schitt]

Ronny wrote:

but there is always going to be less damage to a WLR when truncating without dither than with dither.

Really? I would have thought that would be the other way around.

[Bob Olhsson]

He clearly meant the other way around.

If for no other reason, the fact that the distortion accumulates much faster than the dither does in sibsuquent processing means its always a good idea.

[bblackwood]

Bob Olhsson wrote on Sat, 01 October 2005 14:04

He clearly meant the other way around.

One would hope...

[Ronny]

bblackwood wrote on Sat, 01 October 2005 20:14

Bob Olhsson wrote on Sat, 01 October 2005 14:04 He clearly meant the other way around. One would hope...

Yes, of course, sorry for the typo.

William, I dither from 32 float to 24 bit, if I'm planning on doing more processing with a 24 bit device, if not and I finish processing in 32 bit, I'll dither the reduction straight to 16 bit. I've IDR'd to 24 bit and used TPDF to go to 16 bit and I can hear no difference nor accumulation of dither noise either way. As far as I'm concerned, I've been following the recommendations to dither every word reduction and have found that to work quite well for my apps.

[lucidwaves]

Thanks for correcting me.  I thought that because in a 32 bit float point system it uses a 24 bit mantissa, the dynamic range was the same as in a 24 bit fixed point system.  I guess that there are still the other 8 bits to account for regardless.