I looked at several posts in the forum but still don't have an idea of what the optimum settings are for dither. The default appears to be: Real-time=0 and High Quality=triangle but I can't judge what would work best for a particular project without knowing what effect None,Rectangle and the other settings have on the output quality. Does it matter very little or alot?
Thanks
Optimum Dither settings
Re: Optimum Dither settings
Creating dither requires processing power, so for playing tracks having the dither set to "none" (and the "fast" option) provides the best performance (though with a marginal reduction in sound quality).
When Mixing, Exporting, and other non-real-time processes, the conversion speed is less important, so it is best to set it for "high quality" conversion and apply dither. On most material, "shaped" or "triangle" work well.
Here's the test:
Record something in 32bit format.
Set dither to "triangle"
Export it as a 16bit wav file (test1.wav)
Change the dither to "shaped"
Export it as a 16bit wav file (test2.wav)
Now listen to test1.wav and test2.wav playing through your best quality headphones/loudspeakers.
Can you tell the difference?
Probably not, in which case it does not matter which you choose.
If you can tell the difference, pick the one that you prefer.
I generally use "shaped" (for no particular reason).
Technically - "Triangle" will produce a very low level of "noise" across the full audio spectrum. "Shaped" will produce a slightly higher level of noise, but it is weighted toward the high frequency end of the spectrum (above 15kHz) where it is less noticeable.
In either case, the level of noise is extremely low.
Dither is only relevant when the bit depth is being changed. (It has nothing to do with sample rates).
Internally, Audacity uses 32bit audio, so for the very best sound quality you should work in 32 bit. If you require 16 bit audio (for example if you are going to write to CD), you should still work in 32 bit for the very highest quality, then Export as 16bit as the final step.
If you are simply recording, trimming the ends, then exporting, there is little if any advantage in working in 32 bit - using 16 bit throughout will be fine.
When Mixing, Exporting, and other non-real-time processes, the conversion speed is less important, so it is best to set it for "high quality" conversion and apply dither. On most material, "shaped" or "triangle" work well.
Here's the test:
Record something in 32bit format.
Set dither to "triangle"
Export it as a 16bit wav file (test1.wav)
Change the dither to "shaped"
Export it as a 16bit wav file (test2.wav)
Now listen to test1.wav and test2.wav playing through your best quality headphones/loudspeakers.
Can you tell the difference?
Probably not, in which case it does not matter which you choose.
If you can tell the difference, pick the one that you prefer.
I generally use "shaped" (for no particular reason).
Technically - "Triangle" will produce a very low level of "noise" across the full audio spectrum. "Shaped" will produce a slightly higher level of noise, but it is weighted toward the high frequency end of the spectrum (above 15kHz) where it is less noticeable.
In either case, the level of noise is extremely low.
Dither is only relevant when the bit depth is being changed. (It has nothing to do with sample rates).
Internally, Audacity uses 32bit audio, so for the very best sound quality you should work in 32 bit. If you require 16 bit audio (for example if you are going to write to CD), you should still work in 32 bit for the very highest quality, then Export as 16bit as the final step.
If you are simply recording, trimming the ends, then exporting, there is little if any advantage in working in 32 bit - using 16 bit throughout will be fine.
9/10 questions are answered in the FREQUENTLY ASKED QUESTIONS (FAQ)