How to correctly convert a 24bit master to 16bit?
I've been lurking around here a lot, but now have a question of my own. With a recent project, I decided to mixdown to 24bit/44khz, instead of 16bit/44Khz which is what I have done in the past. After mastering the audio, I was puzzled by this sharp harsh sounding noise i'd get when playing back the file and boosting around the 16khz area.
I understand now that it is an artifact from incorrectly setting the Waves L2 limiter to 16 bit mode in my mastering chain. Presumably this noise was generated by the dither algorithm. I perhaps don't fully understand the concept of dithering however... so correct me if i'm wrong.
The noise at the 16khz area is gone when setting the limiter to 24bit mode. After processing my file through my mastering chain, I then choose to save the resulting file at the CD standard, 16bit/44Khz.
But, is this the correct method to use for converting?
When converting 24-bit to 16-bit, conventional wisdom is you should dither.
When audio files are down-bit-converted, they are truncated, so their waveforms are squared off. Dithering adds a nearly-imperceptible amount of variance to these waveforms, thus taking making them sound subtly smoother and more like your original 24-bit file.
You shouldn't hear any overt noises when dithering. In fact, depending on your monitoring set-up and the style of music mixed, you may not hear the dithering at all. The waveform squaring mentioned above is mostly perceptible in quieter, subtler areas of a track (say, a reverb trail).
Hope this helps.
Thanks for the information on dithering, chrispick.
I've found that the high frequency noise I mentioned earlier is in fact a feature of the L2's noise shaping dither algorithm, where you can push the noise up in to the 16khz area. (Silly me :-)
But, is this a desirable effect? I've run a few commercial tracks through a frequency analyzer and have not noticed this noise shaping effect being used.
Also, am I dithering correctly? I load up my 24-bit file, then I run it through my mastering chain which ends in the Waves L2 set at 16-bit mode. I'm still left with a file at 24-bit, which is where my concern lies. Then I save the resulting file at the CD standard, 16-bit/44Khz.
At the end of your chain, after the L2, you need to record at 16bit with your DAW or cd burner.
It's unusual to say that no noise shaping was done in commercial CDs. Perhaps the ones you chose were mastered to analog tape then reloaded at 16 bit, but that doesn't seem optimal any more. A dithered 24 bit signal can (and in many or most cases will) sound better than a straight analog to 16 bit track.
As for the noise shaping within Waves (IDR), there is a high-frequency peak in their noise, however, it is incredibly LOW in amplitude. It is unlikely that you heard any affects from this on your audio, at least not overt ones. Be sure that dithering is in fact enabled and that it's not just truncating. In general, I find IDR to be a very good and pleasing dither - almost on par with POW-R 3.
OK here's where its messing you up. Don't save to 16 bit!!!
after L2 with 16bit dither, Convert to 16 bit file with no dither. then save the file as "SongName-16bit" or something like that.
1) process with the L2 in the 24 bit environment, BUT with the 16 bit dither
2) convert to 16 bit file, BUT dont use the conversion tool's dither option.
3) Save project.
This is how I was taught. So long as you use the proper dither right before you reduce the bit-depth, you're golden :D