The right way to remove the last 8 bit (few bit) in 24 bit recording

[K8L-64]

I record all my classical stuff in 24 bit.
and my final product is a 16 bit CD.

should I first normalize the 24 bit 2 track master, then re-sample it to 16 bit record?
Or should i re-sample the the unnormalize 2 track to 16 bit first, then doing the normalize process?
Or any other effective method to remove the last few bit in the 24 bit recording? (for the headroom issue, there are always some bit unused)

All I want is removing the unused headroom in the 24 bit recording so as to make a proper master.
Thx for helping~

[surflounge]

Perhaps POW-R dither

24-96 Mastering - Dither

[twwalsh]

you probably don't want to use the term "normalize" around here.

you should do dithering/truncation last. do all your processing at 24 bit. THEN convert to 16 bit. most people prefer to dither for the truncation phase, but some like to argue against it, or that it doesn't matter and so it should be avoided simply on principle.

[Lagerfeldt]

Definitely do add dither when quantizing from 24 to 16 bits, especially if you're doing classical music with lots of dynamics, long tails and natural reverb.

Truncation noise could be real nasty otherwise.

[twwalsh]

Originally Posted by Lagerfeldt

Definitely do add dither when quantizing from 24 to 16 bits, especially if you're doing classical music with lots of dynamics, long tails and natural reverb.

Truncation noise could be real nasty otherwise.

i agree, i like dither almost all the time. however, there have been some nasty arguments to the contrary on this forum.

[Lagerfeldt]

I don't think you'll find many ME's arguing to generally truncate instead of dithering.

And certainly not with classical music.

It's quite simply: people should use their ears. Truncation noise is a lot more audible and nasty than the side effects of a good dither/noise shaping algorithm with most truly dynamic material.

[twwalsh]

Originally Posted by Lagerfeldt

I don't think you'll find many ME's arguing to generally truncate instead of dithering.

one would think so.

however, i direct you to this thread:

Dithering 101

[Lagerfeldt]

Yes, I read that thread.

I don't think you find anyone else than a non-mastering engineer claiming dithering isn't necessary, based on a completely non-scientific "test" which was not properly done even for a simple DIY comparison.