File conversions /vs/ quality?

[hellosailor]

Let's say I'm using 2.x with my files set for no dither, 44k "CD" quality stereo, with the general goal of creating WAV files as masters, and MP3 VBR-2 files for use on noisier portable devices.

And after a while, I get rid of the Audacity files themselves, to save space.

Then I decide, gee, there are some clicks and pops, edits I should do, but what I've got left now are the WAV files.

If I open the WAV file in Audacity (which I assume converts it to AUP format again?), perform my edits, and resave back to WAV format, will I lose anything? Are there going to be "rounding" errors, dithering, any changes created by the extra conversions, or will the data be kept without additional distortions, since I'm sticking to 16-bit and no dithering in all my setup options?

[PGA]

There may be "mathematical differences" but I doubt you will be able to detect them by listening. However, one of the senior elves will no doubt give a detailed answer.

[DVDdoug]

If you open a WAV file and you don't do anything (no dithering or anything), other than save it under another name, the audio data remains unchanged. It gets converted to 32-bit floating-point for internal use in Audacity, but these 32-bit numbers can be converted-back to 16-bits without changes and without errors.

If you make a small change, such as a -1dB volume reduction, save the file, re-open it, and restore the volume, there will be teeny-tiny rounding errors, but these will generally be waaaaaay below audibility. Volume changes (both analog & digital)are an everyday part of audio production & editing, and they are not considered lossy processes, and nobody worries about it, although technically these changes are lossy.

...there are some clicks and pops, edits I should do...

Noise reduction effects & filters can often have unintended side-effects (artifacts). But hopefully, the cure is NOT worse than the disease.

In a scientific/mathematical sense, most effects and filters are "lossy" since they alter the original data (which is of course what you want) and they are not generally 100% reversable so you cannot recover the exact-original bits/bytes.

But again, the intention of these effects & filters is to change the sound.

[hellosailor]

Thanks, Doug. What I've found in practice, is that I can simply "snip" out the click or pop, and the duration is so short that when the music is played there is no perception of anything being missing. I have no idea offhand how small that "snip" is. I suppose I should but...I tired it, it worked, I never looked back.

Anything else, gross corrections to level, etc., I try to make before I'm done with the project and do the exports. But sometimes, I just forget to go on noise patrol.