Exporting, bit rate and dither

[Sk1nZ]

Im wanting to export my track and really confused about dither and bit rate

The track wont be edited again and its ready to upload to various sites, i use ableton 10 and its been made in 24bit / 44100k.

Whats best to do when it will be uploaded to sites like soundcloud, youtube etc.. and what if someone wants to download it and put it onto a cd

Should i have different versions? One for where it will just be streamed and one for where its possible to download and put on CD? and not using dither for streaming and then use dither when it can be downloaded and put on CD?

Or would it be best to use just one export at 24bit with dither?

[whyterabbyt]

Without knowing what those sites are, and also what they specifically do to/with your audio, I think its a safe assumption that they convert to 'their' formats when you upload a master copy. I'd be surprised if they'll even give you the choice of different uploads for different target formats.
Anyways, it makes sense that you use the same master for all sites.

[Sk1nZ]

yeah ive just looked at some sites and they're all different, some only allow 16bit while some allow 24bit. Id like to use just one master but obviously want the best audio quality

What do others do here?

[BertKoor]

Unless you do intros like Roger Waters at -60dBfs, 444.1 kHz & 16bit is the distribution format you should use.

[thecontrolcentre]

What BertKoor said ... 16bit 44kz is what most distributors expect.

[BertKoor]

Forgot to say: dithered, not truncated.

[Sk1nZ]

ok thanks, even if the site allows 24bit? or is the difference so small its not worth bothering about?

[BertKoor]

Do a blind listening test and judge for yourself.

Chances are you have mastered it modern style (single-digit RMS levels)
Even with 80's style mastering (RMS level of -20 dBfs) it was enough to have "only" 16 bits.

Because human hearing has a wide dynamical range, but in a 90dB SPL signal nothing below 40 dB SPL reaches your brain. So the 96dB range of 16 bits suffice quite well.

That's why mp3 compression and the likes (there's better compression algo's used in broadcasting) work transparently.

[heavymetalmixer]

I also recommend you export as 16 bits and dither. IMO, only very dynamic genres like Classical Music and Jazz benefit from the final mixes/masters being at 24 bits.

Btw, I would recommend you to use a meter that shows True Peaks, like Youlean Loudness Meter (which has a free version) so you turn the song down (with fader or gain plugin, not a limiter) and the True Peaks don't clip. Also, because of the lossy compression that platforms and other people will do to the song, the digital peaks and True Peaks will go higher on MP3, AAC, etc, so make sure the True Peaks on the final mix/master don't go above -1.

[mystran]

Should i have different versions? One for where it will just be streamed and one for where its possible to download and put on CD? and not using dither for streaming and then use dither when it can be downloaded and put on CD?

Dither prevents quantisation distortion that happens when you convert floating-point audio into 24-bit or 16-bit format (or really any integer format). The really important thing here is that once the distortion has been introduced, it cannot be removed. Dithering after such distortion has happened just prevents additional distortion from being added, but really at that point it's too late.

So you should always dither every single time you store something into a non-floating-point format. This doesn't just mean your final master, it also means every temporary file you ever create (unless you store them in floating-point). Dithering too many times is basically harmless, where as every time you fail to dither when needed, you risk introducing some distortion. While dithering adds a tiny bit of noise, basic TPDF dither accumulates really slowly, to the point where you would have to dither the same audio hundreds of times before it would ever become an issue.

When in doubt, just dither.

[DJ Warmonger]

Labels always ask me for 44100 Hz / 16 bit. Then they distribute it across many shops and services automatically, each probably does its own conversion. Production technique is much more important than any audio transcoding.

Higher bit rates or sampling rates make sense only if you're at immediate stage and plan to do some more processing over finished tracks, i.e mastering. Personally I work at 88200 Hz to combat aliasing, but then downsample for distribution.