Converting Sample Rate /Dithering

[ukoomrann]

I am about to master a cd recorded in 24 bit in 96khz .
what should be the right order?
First convert to 44.1 and then dither to 16 bit
Or first dither to 16 bit and then convert to 44.1
As far as I know the dithering should be absolutily the last thing done

Thanks

[minister]

right. dither last. convert to 44.1, then dither to 16.

[aux]

Minister is right: downsample to 44.1, dual mono files, 24 bit and the same file format. Then dither/noiseshape down to 16 in a new session.

[Matt_G]

Assuming that you are finished doing any mastering processes at 24bit 96kHz, definitely do your SRC to 44.1 at 24bits, better still use 24bit TPDF dither during the SRC from 96k->44.1k because the SRC process actually increases the wordlength above 24bits & for the purist results, it needs to be dithered back to 24bits. It's also wise to leave yourself a little bit of headroom with the output of your limiter (-0.3dbfs) so that your SRC has a little room to breath. Once you have your converted 24bit 44.1k file then use a noise shaped dither to reduce the wordlength down to 16 bits. Something like POW-r type 2 or type 3 is perfect for this purpose.

Personally I use Barbabatch for SRC as it can do the 24bit dither during the conversion process & sounds superior to Protools SRC.

Matt