Hi, when i’m only performing simple cutting/trimming edits, or exporting a long file as multiple tracks, or i’m just exporting a wav to mp3, I read it’s safe to turn dither off. I’m unsure if I have to change the project settings to 16-bit first though? If so does it matter if I set it to 16-bit just before I export or does it have to be set at the start of the project? It’s currently set to 32-bit float and I set dither to shaped for all edits involving filters/processing.
As long as the input and output formats match and all you do is cuts, dithering is not needed. The 16-bit > 32-floating > 16-bit transition has no in-between values to resolve. The instant you do effects or change levels, you could (easily) have 32-floating sound values inside Audacity that do not translate directly to 16-bit. Now you got problems.
The default dither system in the latest (I think) two Audacity versions is inaudible in anything but laboratory inspection. And yes, Audacity is Scientist-Hostile because of that. “How come the ultra-precise sound values I put in didn’t come out?!?”
Audacity may not be the program for you.
Thanks for your reply, so when I’m just doing cuts I can turn dither off and I won’t need to set to 16-bit. Correct me if i’m wrong. The only reason I turn dither off is because I learnt the hard way after saving and importing multiple times the hiss became audible and I found out dither was the reason, so now I save the project rather than exporting. I complete a project, master it in other software then use Audacity to convert to mp3 or to just split to tracks so i’d rather not apply dither multiple times. I didn’t realise the dither system in the last two versions is inaudible. Thanks
There was a bug in some old versions of Audacity that would cause excessive dither noise on stereo tracks when reducing the bit format from 32 bit float to 16 bit. That bug has been fixed. However, dither noise will still increase if applied multiple times by about 1 dB per conversion.
Ideally, conversion from 32 bit float to 16 bit should be done once only and should be dithered.
Not that the sound quality loss involved in MP3 encoding multiple times is MUCH greater than the sound quality loss through applying dither multiple times. Dither applied 10 times is likely to be just about audible during extremely quiet parts of a recoding (such as at the end of a fade out), whereas MP3 encoding applied 10 times is likely to sound dreadful.
The upshot is that if you will be wanting to work on a track again in the future, keep a backup copy in as high quality format as is practical. DON’T rely on an MP3 as your backup. Ideally (from a quality point of view) is to use a “32 bit float” format WAV file. The size of a “32 bit float” format WAV file is double that of a 16 bit WAV file, but is likely to be smaller than saving an Audacity project, and you don’t need to worry about the AUP and _data folders becoming separated because a WAV file is self contained.
32 bit float format WAV is exactly the same sound quality as the Audacity project - zero quality loss.
WAV files may be “zip’d” to reduce their size a bit. ZIP archiving is lossless.
In short, my recommendation is to leave dither on (set to “shaped”) and if you want a “perfect” quality backup, export as “32 bit float WAV”.
To export as 32 bit float WAV, select “other uncompressed formats” as the file format, and click on the “options” button to set “WAV” as the header and “32 bit float” as the format: http://manual.audacityteam.org/o/man/file_export_dialog.html