Dithering

Has anyone ever heard a difference between dithered and non-dithered audio? I always dither at the end but I don't hear a difference. Thoughts?

All i know is that Modern Talking wouldn’t sound the same without Dieter
 

I do because you're supposed to, but honestly I can't tell the difference.

I think the idea is that a digital waveform will never be perfect, so dithering would improve it mathematically even if you don't hear it through your system. It just smooths out any quantization error.

If in doubt ask Dan Worrall (despite the title it's more a why/what's the audible difference than a what)

  On 3/6/2020 at 11:50 PM, mcbpete said:

If in doubt ask Dan Worrall

 

Expand  

This is the best video on this topic I've seen so far

Dithering in unobtrusive, low on CPU, and mandatory. As strange and exaggerated as it may seem, it can make an audible difference even when going from 32/64 to 24bit. Truncation is an unpredictable bitch.

The various Airwindows options are great, especially all the variations around the NotJustAnotherDither algo. AW Monitoring lives on my 2buss while producing / mixing, and of course at export. Beware that your DAC isn't supposed to dither what it outputs either, so it definitely makes sense to monitor through it.

Ps: I've also heard some significant difference with low quality streaming à la Soundcloud.

  On 3/7/2020 at 11:53 AM, Nil said:

Dithering in unobtrusive, low on CPU, and mandatory. As strange and exaggerated as it may seem, it can make an audible difference even when going from 32/64 to 24bit. Truncation is an unpredictable bitch.

The various Airwindows options are great, especially all the variations around the NotJustAnotherDither algo. AW Monitoring lives on my 2buss while producing / mixing, and of course at export. Beware that your DAC isn't supposed to dither what it outputs either, so it definitely makes sense to monitor through it.

Ps: I've also heard some significant difference with low quality streaming à la Soundcloud.

Expand  

Thanks for this response, although I think a lot of low quality sounding Soundcloud uploads come from unexpected level peaks caused by rounding errors that occur during the file compression to a lossy format which can be fixed by using a ceiling level of below 0db, some say -0.3db is enough, some say -1db is enough, depends on the kind of limiter you use I guess (true peak/non-true-peak). But if you're telling me that the absence of dithering or wrong dithering is a factor that contributes to that problem I'm gonna believe you

Absolutely. It's a deteriorating factor, adding of course to the super lo-fi, brutal Soundcloud conversion.

On a related note, I for one can't wait for Soundcloud to adopt some sort of loudness normalisation. I do more and more mastering, and most of the time Spotify is the target for the client : dynamic masters sound so much better, every time, and streaming makes no exception there.

I'm getting used to offer 2 versions because of SC : a balanced, dynamic one with a healthy crest factor that sounds absolutely great on its own on all streaming platforms, on wav as well of course, whenever played live on a nice PA etc... and a stupidly limited one for SC so that it can "rival" with all the absurdly loud, crushed tracks up there.

Ps: have you noticed the newly added HD icon whenever uploading a lossless track to SC though ? The streaming fidelity seems a bit better now with such tracks.