Dithering

[inmatus]

Hi
Just want to check if I understood this right

I deliver my music to my distributer as wav and can do so in 24 bits

So when mixing is done I bounce to 24 bit WAV -no dithering needed.
Then I master it and since I deliver in 24 -no need for dithering either

I also upload songs to places like soundcloud, i.e. mp3. I guess no need to go 24->16 but encode from the 24 bit file. In other words no need for dithering either

Correct?

[marcuswitt]

It depends on which distribution media you're delivering your music for, I'd say.

Nowadays, in times of online distribution and other high-res media (let's put the mp3, AAC, etc. formats aside for a moment) the necessity for extra-dithering is not as big as it used to be, let's say, 10 to 15 years ago. So, I guess you're doing nothing wrong with submitting your music in 24bits resolution. But, as you surely know, if you want to spread your creative outcomes via CD Audio then you can't avoid to apply some dithering (24 to 16 bits) to your audio material, generally speaking. But even then it depends on the source material wether you can simple 'truncate' from 24bits down to 16bits without any audible loss of quality or if it's really necessary to use a dithering algorithm to reach a better quality. The last time I've delivered a dithered 16bits version to a customer was in 2007. Since then I print my mixes out in 24bits only.

[inmatus]

Thank you
The distribution I am using is only for digital services like spotify, iTunes etc etc. After I delivered the wavs it is out of my hands. Wimp (Tidal) have high quality stream that is said to be "CD quality".
If that means 44,1 /16 bits I dont know. I think the other high q streaming like Spotify use FLAC?
Like I said, my distributor, Phonofile, accepts 24 bit WAVs. I recon it is better to deliver in that than in 16 bit when whoever digital service then encode it further?
The higher Q of the audio to start with prior encoding to whatever format should give a better result?

[ScuzzyEye]

There's no reason you can't add 16-bit dither to a 24-bit export (well, when using Ozone, as Reason doesn't let you do this). It just raises the noise floor to -96 dB. Then if it is truncated to 16-bit there will be less quantization noise.

If you're not aware what dithering is trying to accomplish, it's this: Just as there's aliasing artifacts when you try to store a frequency too high for the sampling rate to represent, there's quantization noise that occurs when repeating signal hits exactly the same level cycle after cycle. It can produce a sound that's unrelated to the signal, and is not pleasing to the ear. The idea of dither is to add a type of noise that the ear doesn't find offensive that jostles the levels slightly. This way even if the input signal hits the same level, the added dither will make it fluctuate slightly. It's replacing one type of noise with another.

Because dither is designed to be inoffensive, and not intruding you can add 16-bit to a 24-bit file, and no one is likely to complain. If you have anything with a live recording (i.e. not all soft synths) you likely already have a noise floor above -96 dB, so it will be inaudible.

Oh, almost as an after-thought: Some dithering programs are aware of the sampling rate, and if using shaped dither will take advantage of the extended bandwidth and place the majority of the dither energy up higher. The may be a problem if you're exporting at a higher rate, and expecting to re-sample lower later when those higher frequencies are removed.

[inmatus]

I just have a bit hard to really understand it Cant help it

It is about to introduce noise to mask the quantization noise that is introduced when a moving to a lower bit rate. So.. if I bounce 24 bit and deliver in 24 bit there is no export to lower bit rate and no need for dithering

When that 24 bit is encoded to mp3 or flac etc- I guess it is pointless to talk about dithering since the ecoding process is handling it in its algorithms

What is the 24 bit dithering that is possible to use with Ozone? Reason can save in max 24 bits so what is the need for that?

[ScuzzyEye]

What is the 24 bit dithering that is possible to use with Ozone? Reason can save in max 24 bits so what is the need for that?

Reason uses 32-bit floating point internally. So there may be a reduction when going to 24-bit. Plus it's not just about bit-depth reduction, but hiding the quantization noise at what ever level is being used. So yeah, you can apply 24-bit dithering to a 24-bit export. It'll have the dither noise at around -144 dB.

[inmatus]

Well.. If that would be a dithering case going from 32 bit floating to 24 bits.. how come Reason dont support dithering for this?

I asked Props support if dithering when bouncing to 24 bit was needed and they said no?

[ScuzzyEye]

Well.. If that would be a dithering case going from 32 bit floating to 24 bits.. how come Reason dont support dithering for this?

I asked Props support if dithering when bouncing to 24 bit was needed and they said no?

It's arguable that dither isn't needed at 16 bits for anything but test tones, and maybe very quiet material. Once you have musical content quantization errors get buried in it. So yeah, it almost certainly isn't need when you're quantizing into 16,777,216 possible levels. But if you're a guy who likes to wear both a belt and bracers, just to be safe, then 24-bit dithering isn't going to hurt anything.

[inmatus]

I think I got it now
Thank you all