First of all, I disabled all Dithering options from settings. So I import an audio file (Let’s say it’s Bit Depth is 16 Bit) and the very first thing I do is right click the track > Format > 16 Bit PCM (change the Bit Depth from 32 Bit Float to 16 Bit to match the input file), then edit it (Normalizing, Looping, Fading In/Out and/or Adding/Removing Silence is what I refer to when saying “Edit the audio” ) and finally Export the audio as FLAC with the Bit Depth that matches the input file (In this case 16 Bit).
My question is: Will the exported file be lossless compared to the original or there will be Rounding Errors because of a previous step I made when Editing or Changing the Bit Depth? and if there are rounding errors or any sort of quality loss, is there a way to completely avoid Dithering, but also to export without errors or loss of quality as FLAC with the Bit Depth matching the input file?
There will be rounding errors when applying a fade.
Imagine for example that there is sample half way along a track that has a 16-bit value of exactly 21 (as a 16-bit number). You then apply a fade out to the entire track, and because our chosen sample is half way along the fade, its value should be halved. That means that the new sample value should be 10.5, but the16-bit value must be an integer (a whole number), so it is impossible to represent 10.5 exactly as a 16-bit integer.
For best quality in this case, leave the track format as 32-bit float and use “Shaped dither” on export. Exporting as 24-bit FLAC will be slightly more accurate. Exporting as 32-bit float WAV would be the most accurate.
For normal music audio, it is unlikely that you will be able to hear any difference whether using 16 / 24 /32 bit with or without dither enabled. We are considering very, very tiny differences.