MOTUnation.com

I am digging back through archival projects that are being revisited for overdubs and remixes, and many of them have 16-bit sample format due to my inattentiveness during my first month or two of using DP 3.0 and not noticing that my preference for 24-bit recording had gotten overridden in some cases.

Should I just use the sound bites window to concvert them all to 24-bits, or are there dithering issues going that direction as well as when going from 24-bits to 16-bits? Can I assume it simply tacks on an extra eight bits?

If there are quality issues in doing this conversion, am I better off doing it instead outside DP? I just installed BIAS Peak LE last night and could use it to do the job, but it's more work to do the multiple imports/exports from DP soundbite files than to simply do it destructively within DP.

I also recorded many of the original sessions at 96k, until later on backing down to 48k and/or 44.1k for most pop/rock work due to audio overload issues on my iMac G4 and also issues with using external S/PDIF connected gear for audio processing.

The material in question is unlikely to suffer from sample rate conversion as it is electric rock music, but I figure if I go to 24 bits first, I will get better results, and also should probably stick to multiples of two (i.e. go to 48k vs. 44.1k).

For this latter task, I understand that either Ozone or BIAS Peak LE may give me better results than DP, which is the only reason I mention it here (as there are other topics devoted to sample rate conversion), as I may as well do the whole task (including sample format conversion) in BIAS Peak LE (with or without the Ozone plug-in) if I need to use it for the sample rate conversion anyway.

You won't lose any quality going up to 24 bits but you may lose a lot of time and disk space. Doing this will only be an advantage if you're overdubbing new parts that you might actually hear a difference on, like the lead vocal. If not, I'd say stay in 16bit for those projects.

Yes, the main goal is to re-track ALL of the vocals, I'm just re-tracking the bass and adding a few additional background synth and organ pads before handing the files over to the vocalist to redo on his own time. He was very inexperienced when we did the originals, and very intimidated by the whole studio thing. So I agreed to give him revised audio tracks for him to import into ProTools and then record his new vocals and export those for me to integrate back into the final mix. I DEFINITELY want that to be 24-bits!

I'll be using the TC Electronic Triple-C for compression, and the Kurzweil Rumour for reverb. In order to stay in the digital domain, I will be using S/PDIF at either 48 kHz or 44.1 kHz sample rate (my recollection is that S/PDIF is limited to 48 kHz max, regardless of whether the manufacturer is willing to put a bit of extra money into supporting higher rates).

Convert to 24 bit. Any processing including simple level shifting and panning for a mix will bring better results, with more detail in the quiet passages. Reverb tails are greatly improved. Add new tracks at 24 bit, add plug ins and FX and mix to 24 bit. Then master to 16 bit with a peak limiter plug that has dither and noise shaping. This seems to produce the best sounding final result.

The only time not to go to 24 bit is if you are not going to alter the data in any way. A simple copy of a 16 bit file will be made noisier with no additional information if you add the zeros and then dither.

As for the higher sample rate material convert to 24 bit first, then sample rate convert, then dither down to 16 bit.

yep, I say upsample. Even changing the gain of a track can produce quantization errors...if you convert to 24 bit, you are providing yourself with the digital headroom (note: not traditional audio headroom in this case) for those errors to occur within.

Check out Bob Katz's book. It is extremely helpful on these subjects and is one of the most educating books I can recommend.

Thank you both for very helpful and informative posts.