It was my understanding that repeated edits of WAV files are lossless. But I read somewhere that there can be loss if the dither is set wrong. Is this true?
I’m also wondering whether there is a lossless format with smaller file size than WAV.
By default Audacity uses “32 bit float” format. Editing in 32 bit float format is completely lossless.
When exporting as 16 bit WAV (a “normal” WAV file), there is a very tiny amount of loss due to the conversion from 32 down to 16 bit because 16 bit is not as precise as 32 bit. That is not normally a problem because 16 bit is still very high quality (CD quality). However, if you open a WAV file, do some edits, export as 16 bit WAV, close the project, open a new project, import the WAV file, do some edits, export as 16 bit, close the project, open a new project, import the WAV file… over and over again, that tiny bit of loss occurs every time that you export, and eventually it will become noticeable.
In most situations the loss that we are talking about is so small that it is insignificant.
While working in the Audacity project, provided that you are using the default 32 bit float format (see the info in the panel on the left end of the track) you can keep making as many edits as you like with no loss of sound quality.
You could think of this like editing a photograph and working in extremely high resolution (thousands of dots per inch), then saving the picture in 600 dots per inch. 600 dpi is still very high quality, but not quite as high as what you were working with.
“FLAC”. The file size is usually about half the size of a WAV file but the same quality as WAV.
“While working in the Audacity project, provided that you are using the default 32 bit float format (see the info in the panel on the left end of the track) you can keep making as many edits as you like with no loss of sound quality.”
Do you mean : if you don’t export and re-open to edit?
What if the default is set to 16 bit?
What about the dither issue?
How does FLAC compare with regard to repeated edits?
The reason I have hesitated to get the new Audacity version is because I feared there would be a loss of settings, and differences to get used to, which may slow me down unless I’m prepared for them. Is this the case?
Yes, I mean that while the format is 32 bit float (as it is in an Audacity project), then edits are totally lossless. Thus, if you are wanting to make some edits now, and then more edits tomorrow, it is best to save the project after today’s edits, then open the project tomorrow and make the other edits. Audacity projects are not very good for long term ‘archiving’ (they are very big, and you have to take great care to keep the AUP file and its data together), but they are very good when working on a project over a period of days (you still need to take care to keep the project file and all its data together - see: Audacity Manual)
Not good.
16 bit quality is fine for the finished product, but for processing the audio you really need higher quality so as to avoid “quantize errors” (rounding errors in the number crunching). Also, Audacity does not handle 16 bit tracks very well making 16 bit processing a little less good than it (theoretically) should be.
Dither is not required for 32 bit float format, so when working in 32 bit float format, it’s not an issue. The ideal work-flow is to work in 32 bit float format throughout, then export as 16 bit WAV (or FLAC format) when the job is complete. For optimal sound quality, dither should normally be applied when exporting to 16 bit, but ideally that should only happen once (at the end when you have finished the job).
It behaves exactly the same as WAV format.
if you installed 2.0.3 using the recommended installer, then 2.0.6 can be installed “over the top” of the old version. That will update Audacity to 2.0.6 and your existing preferences are retained.
There aren’t really any radical differences - mostly bug fixes and some worthwhile refinements to various features. There are a few changes to the menu layout, but nothing that you won’t get used to pretty quick.
Generally it is best to leave dither set at the default settings:
Real-time Sample Rate Converter: Dither = None
High-quality Sample Rate Converter: Dither = Shaped.
What are the exact circumstances that lead you to think that “none” would be better?
“What are the exact circumstances that lead you to think that “none” would be better?”
I don’t know anything about it. I don’t really understand dither. I wanted to clarify your statement that:
“Dither is not required for 32 bit float format”
I’ve been using Download Helper to download music videos to convert to WAV for a specialized music show. But just recently Youtube seems to have again succeeded in blocking DH, so I must record in real time.
Dither is applied when converting from a high resolution format to a lower resolution format, such as from 32bit to 16 bit.
If you are working entirely in 32 bit float, then there is no conversion taking place, so dither is not applied.
When you make the final export to 16 bit WAV, there is a conversion to a lower resolution format, so dither should be applied.
There is a detailed article on the Audacity wiki about dither (Missing features - Audacity Support), but the short version is that the default settings are generally best (which is why they are the defaults )