16 bit to 24 bit conversion before mastering?

  • Thread starter Thread starter vogler
  • Start date Start date
V

vogler

New member
hello,
a friend of mine gave me CD to "master" (home...) it. Surely it's 16/44.1.
Should I convert it to 24 bit and upsample to 96 kHz before loading to software DAW? Of course my DAW is set to work @ 24/96. Or should I only dither it while exporting back (to 16/44.1) from DAW after the job is done?

My another question is: when I recorded into DAW @ 24/96 and I want to export this single recorded file to 24/96 without doing any chages on it - should I dither it?
Thanks for help,
/vogler
 
Upsampling will do you no good. The added information is already gone, you won't be able to put it back by upsampling.

There is no reason to convert it to 44.1k/16bit before mastering, that is part of the mastering process. Mix it down to 24/96 wav files and put them on a data CD, like you would word documents or any other type of computer file. Take that to go get mastered.

The mastering engineer is the one that should be dithering after he does all the processing. Dithering is the final step just before the master CD is burned.

BTW, most pro engineers record at the target sample rate (44.1k for CD, 48k for video), there really isn't any compelling reason to use the HD space and computer resources just to try to record frequencies that most mics can't pick up, most playback systems can't reproduce and most people can't hear.
 
Farview said:
There is no reason to convert it to 44.1k/16bit before mastering, that is part of the mastering process. Mix it down to 24/96 wav files and put them on a data CD,
Click to expand...

thanks, but if it's me who is going to "master" it - should I convert the 16-bit files to 24 bit BEFORE l start working on them in my DAW (w/ working quality set to 24 bit) or it's not necessary? unfortunately someone mixed down to 16 bits before mastering.
regards,
/vogler
 
There is no point to upsampling or changing the bit depth after it has already been resampled down. The information is already lost.

Most DAWs will process everything at 32 bit float regardless of what the original file was. The DAW would have to be really out of date not to.

What DAW are you using?
 
vogler said:
Surely Sonar will handle with 16 bits.
Click to expand...
Everything handles 16 bit. It was a question of whether or not it does calculations at 32 bit float. If you have the newest version of Sonar, you might be running at 64 bit.

vogler said:
could you tell me in what situation i do NOT add dither while exporting projects?
Click to expand...
I only add dither as I am burning the CD. I keep everything at the highest bit depth until the very end when I have to dither down to 16 bit for the CD.
 
Actually it's not bad idea to both upsample and increase bit depth if you plan on processing the files further. Upsampling has a tendency of helping to smooth out the upper end when using limiters, while all processing benefits with an increase wordlength. Even though the original low level information is lost, processing adds cummulative noise and quantization distortion that is better left at lower levels until dithering and going back to 16 bit.
 
All the math through the signal chain is going to be done in 32-bit or 64-bit float regardless of whether you start with a file in 16-bit or 24-bit integer format. Thus, you aren't getting any additional precision in those calculations by starting with a 24-bit file scaled up from a 16-bit file; whether you start with a 16-bit sample or a 24-bit sample converted from that value (which is just the 16-bit sample with eight zero bits tacked on), the resulting float value used throughout your effects chain should be identical, bugs notwithstanding.

In short, unless you use a different app to do the conversion that uses a different algorithm for increasing the bit depth, all you're really achieving by bumping it up to a higher bit depth ahead of time is wasting space on your hard drive.

Upsampling to a higher frequency would be a good idea, though, if you plan to use any FFT-based effects. IIRC, higher sampling rate generally translates into more FFT buckets (assuming a fixed duration FFT rather than a fixed sample count) and thus you can potentially get greater precision on things like pitch detection, etc. even with sloppier algorithms.
 
dgatwood said:
All the math through the signal chain is going to be done in 32-bit or 64-bit float regardless of whether you start with a file in 16-bit or 24-bit integer format. Thus, you aren't getting any additional precision in those calculations by starting with a 24-bit file scaled up from a 16-bit file; whether you start with a 16-bit sample or a 24-bit sample converted from that value (which is just the 16-bit sample with eight zero bits tacked on), the resulting float value used throughout your effects chain should be identical, bugs notwithstanding.
Click to expand...

This depends on the wordlength used for "bus" within the DAW. With some DAWs when you setup a session for 16 bit while all processing may be done at a higher bit depth within the plug internally, the result gets truncated back to a 16 bit bus when output from the plug-in creating cumulative truncations. If possible it's a good general rule of thumb to remain at the highest bit depth for as long possible before dithering down to 16 bit.

dgatwood said:
Upsampling to a higher frequency would be a good idea, though, if you plan to use any FFT-based effects. IIRC, higher sampling rate generally translates into more FFT buckets (assuming a fixed duration FFT rather than a fixed sample count) and thus you can potentially get greater precision on things like pitch detection, etc. even with sloppier algorithms.
Click to expand...

Upsampling for non-linear processing allows the algorithm to push artifacts to a higher frequency range that can then be removed by filtering the artifacts above Nyquist. Many plugs have upsampling built into them for this reason, for example PSP Xenon and various upsampling EQs.
 
Let's pose the question of high quality plugins, dsp, items.
There are certain reverbs and compressors that can utilize 96k, doing so tends to make them sound for "natural"
 
You must log in or register to reply here.
Back
Top