Dither?

[mml]

OK, so I know this is in Bob Katz' book, but I don't have it with me and I need to bounce some mixes down. My question: Do I need to put dither on the Master buss if I'm going from 24 bit 44kHz to 24 bit 44kHz? Does it help correct the summing math distortion? This is pre- master. I'm just trying to get my mixes to the highest quality stereo files they can be, before mastering. It will be dithered at that point again of course, but is there an advantage to adding dither when mixing any tracks together?

Thanks for any enlightenment.

[trilliumsound]

Dithering should only be applied when reducing wordlenght.

Cheers.

Richard

[Jamz0r]

Right.
Don't dither until the wavs are being converted to 44.1/16bit masters.

[MichaelPatrick]

I agree with the comments above.

Leave both the sample rate and the wordlength at their highest all the way through mixing and post production. Send that into the mastering process.

In (pre)mastering you downsample (e.g. from 96 to 44.1) and the LAST thing you do is reduce the wordlength. Wordlength (bit-depth) reduction is the only step requiring dither.

Dithering is like circumcision. Never do it twice!

[16/44.1]

Dithering is like circumcision. Never do it twice!

Why not twice?
During A/D conversion there will be dither added.
During conversion from 1 bit bitstream -> 16 or 24.
During mixing/mastering.
More perhaps.
Dither is just random noise..........
It will also be added during the transfer from analogue tape to digital.

[Jamz0r]

O_o

[MichaelPatrick]

Fair enough.

I was trying to be clever while thinking about why I dither only once during a downsample w/bitdepth conversion. There are other conversion scenarios.

[16/44.1]

Originally Posted by MichaelPatrick

Fair enough.

I was trying to be clever while thinking about why I dither only once during a downsample w/bitdepth conversion. There are other conversion scenarios.

[16/44.1]

But the less you do with the audiodata, the better it is.

Recording in 16 bits or 24 bits.
Dither (noise) will be added, no matter what........
Editing, mixing, eq'ing, etc in the 24/32 bit domain.
Round off (dither) to 16 or 24 bits for permanent storage (depends of the medium you are using).
Preferably 24 bits.

Mastering is of later concern.......

[Cellotron]

Originally Posted by 16/44.1

Why not twice?
During A/D conversion there will be dither added.

With the majority of current converters this is not true at all. For example my Mytek ADC gives the option of recording at 24bits without any dither added, or at 16bits with either TPDF or Mytek's own "Supershaper" algorithm. If you are recording at 24bits then there is no need to add dither during AD conversion at all, as the 24th least significant bit will be far below the noise floor of your source anyway.

Best regards,
Steve Berson

[trilliumsound]

Originally Posted by Cellotron

With the majority of current converters this is not true at all. For example my Mytek ADC gives the option of recording at 24bits without any dither added, or at 16bits with either TPDF or Mytek's own "Supershaper" algorithm. If you are recording at 24bits then there is no need to add dither during AD conversion at all, as the 24th least significant bit will be far below the noise floor of your source anyway.

Best regards,
Steve Berson

I second that.

[16/44.1]

Originally Posted by trilliumsound

I second that.

For 24 bit you're correct..............i hope.
Besides that, if the noisefloor is way above the 24 bit lsb, what's the benefit to record at 24 bits?
Is 16 bit in that case not more economical?
My noisefloor is around or just above the 16th bit.
The noisefloor of my pcm is on the 16th but most likely at the 15th bit.
I can see that on my tv monitor.
So i'm still converting it with the old Sony processors.
That data will go straight to DAT/CDR.
And/Or it can go to the FC1/PC to convert it to 24 bits.

But how about the noise added at the analogue stage prior 24 bits conversion?