Audiobus: Use your music apps together.
What is Audiobus? — Audiobus is an award-winning music app for iPhone and iPad which lets you use your other music apps together. Chain effects on your favourite synth, run the output of apps or Audio Units into an app like GarageBand or Loopy, or select a different audio interface output for each app. Route MIDI between apps — drive a synth from a MIDI sequencer, or add an arpeggiator to your MIDI keyboard — or sync with your external MIDI gear. And control your entire setup from a MIDI controller.
Download on the App StoreAudiobus is the app that makes the rest of your setup better.
Where can I apply dithering in this workflow?
I created a project in AUM, passed it to AudioShare and upload to SoundCloud. The soundcloud result is
noisy while the original was not so I think I need to dither before uploading. What app or process should I use to dither the soundiile before uploading to SoundCloud and getting tasing the noise floor due to a change in bitrates. I'm basing the question on this video's advice:
Comments
According to another thread I can use DDMF NoLimit FX to dither. I'll just stick one in front of the record channel in AUM and see what I get compared to the recording I uploaded that sounds really bad. Stay tuned.
The video gives a clear advice: only dither exactly 1 time at the very final stage of the mix.
If you dither and decide to continue working on the track, then the noise added by the dither process will degrade your signal.
But since dither is a feature applied to the lowest bits only, it's generally hard to be detected at all during (regular) listening.
(you'd have to amplify the silent parts of the track significantly to notice the differenc)
Your upload is always processed by a special Souncloud codec (128kbit mp3), which may introduce some noise, but it's more likely your track's volume is too low or there's content that drives the mp3 encoder nuts.
Do the upload codecs like soundcloud dither ? And if you have a dithered wav, can that affect the mp3 conversion ?
I have no idea what the Soundcloud encoder does internally, but it's a magnitude above whatever dithering is about.
(I once uploaded an almost bitingly sharp microphone recording that came out rather pleasant after SC's codec did it's job)
@McD A blog post by the same author specifically on dithering, which I found useful:
http://productionadvice.co.uk/when-in-doubt-dither/
Dithering will do nothing for your issue, sorry. Here’s some more info on it:
https://forum.ableton.com/viewtopic.php?f=1&t=214124
Ta
Thanks all for your input.
At least I learned how to dither and what tool will work.
My project still sounds like crap. I need to re-make the project in AUM.
I created a bass with a lot of sub-oscillating bass but I think my volumes were just too
hot. It sounded great while making the recording but I was tired and it was loud.
I need to insert a limiter on the recording channel too I think to avoid such a hot mess
in future.
If you want to hear the effect of using 24-bit Dithering before uploading to SoundCloud
compare these 2 tracks.
FYI: The whale choir starts at the 2 minute mark. I used an FX App to create it. Any guesses?
I can't hear much of a difference except for maybe reduced volume in the dithered track
which I added to avoid the hot mess again but it's based on trying to process with dithering an already bad wave file that is beyond repair.
Dither noise is almost always way, WAY below the noisefloor of most pro-audio equipment. I'd bet 99.99% of most professional audio engineers can't even hear dither unless it's in very specific cases meant to highlight it, and at volumes much louder than normal playback on high end audio systems. While it serves a useful purpose, it's effect on audio production is about as minimal as it gets.
The only reason musicians these days even know about it is because early on some manufacturers were trying to over-sell their mastering plug ins and gave an option of dither types to people. Otherwise it's something that ideally should just be done when needed by the software, without any knowledge or input from musicians in the first place.
Can I copy this to the FAQ tread and give you credit for the text answer? It might save someone else the time and money to try dithering for no benefit.
Sure.
Thanks. It might be a few days but I'll get to it.
@Tarekith Thanks !! Wanted to say exactly same points like you, you saved my time :-) Holy truth !
Here's another example? Question?:
I converted thousands of 24-bit wav files to 16-bit. Sometime during that job, I forgot to enable dithering. The 24-bit source wav files were of a loud very bright FM synth, and a few of the resulting non-dithered 16-bit wav files were 'scratchy' and sounded like I ran it through a bit crusher. But only a few of the non-dithered 16-bit files sounded like that. Most sounded fine.
Was that 'quantisation distortion' that made it sound 'scratchy'? (the source files were loud)
Why did only a few of the files sound like that? (they were all from a loud bright FM synth)
I believe I used foobar2000 for the bulk conversions.
Thanks.
@ocelot result of FM synthesis can contain a lot of very hig frequencies above “nyquist frequency” (half of sample rate) - which then causes aliasing ...
normally this problem is solved by oversampling in synthesis phase..
maybe that dithering removed those aliasing artifacts ?
But this is more about sample rates than bit rates - what was source/target file sample rate, was there some downsampling applied ?
can you provide examples of source / target files, and target eventually both with and without dithering ? ideally in wav format...
Thanks @dendy . Source files were 24-bit 44100-Hz wav, target was 16-bit 44100-Hz wav. Only bit depth reduction, no sample rate reduction or other processing on files.
interesting.. still think it has something to do with frequencies on the edge of nyquist frequency... would be interesting to hear those files and watch them on spectrum analyser
@ocelot I think that if you hear the difference, it makes a case for using dither. I also agree on all points with @Tarekith, who is by all accounts way more kowledgable than me 😊 But he makes the point that software should handle this, if neccesary, without user input, which might not be the case...
It seems to be a sensitive topic, though, and I really have no stake in the discussion 😇
Just so I'm not inadvertently spreading misinformation. You should definitely be using dither if you have the option to do so when converting to a lower bit depth. My point about the software doing it automatically for you is more of a "in a perfect world musicians wouldn't have to worry about it and it would just be done for you when appropriate" kind of statement.
Sadly, these days in reality it's still something that is left up to us to choose to use or not. And it serves a very valid and useful purpose, even if the effect is incredibly minuscule in the long run.
Hope that clears things up, and I'd appreciate if this was added to the FAQ since it sounds like I'm advocating people not to use dither there.
Thanks!
I'll add the additional advice to the FAQ. Thanks for stating the case when it might be useful. Uploading to streaming service is not such a case.
Thanks!
Thank you for sharing your experiences and insights and putting your name on it.
Thanks @denx it was noticeable. Later when listening to the converted files, a few sounded awful (scratchy like a bit crusher) so I backtracked and the only issue I discovered was that I forgot to turn on dithering.
I've tested different types of dithering in various software, and it's usually not at all noticeable to my ears. But in the case I mention above, it was nasty. But it could have been something completely unrelated like a memory glitch that occasionally happens when bulk-converting to a different disk.
Thanks again to all.
good hint @Max23 , but it's not extreme - just perfectly in accordance with math:
8 bits cover a dynamic range of 8 times 6 dB = 48 dB
the lowest bit represents -48 dB, so the quantization noise is in that range.
with 12 bits you'd see 12 times 6 dB = 72 dB, which results in -72 dB
same calculation with 16 bits gives -96 dB, that's why it isn't even perceivable on most gear
this simplified calculation is based on theory assumptions to illustrate the principle
real world results will be at least 6dB 'worse', but 12dB wouldn't be a surprise either.
It depends on the gear used, not all circuits are created equal...