I have a set of FLAC files of a live concert digitized from an old cassette (which has since been destroyed). The FLACs are split into 13 seamless files, one per song. The tape was running slow when digitizing, and now I want to speed it up in Audacity.
I’ve found the exact pitch/speed the files should be, but I’m not sure if from an archival/lossless point of view it would it be better to join all the FLACs together again before pitch-shifting and then split them back apart (a lot of work), or would it be ok just applying the pitch-shift to each FLAC file as it stands?
Eg, from a really geeky technical point of view does pitch shifting involve some sort of rounding up/down interpolation which could render the beginning/end of a file different than if that section was in place within the original recording?
(I observe this in Photoshop sometimes if I crop a picture then scale it up/down the outer pixel edge is sometimes interpolated differently than if I scale the picture first, then crop it. If that makes sense?)
I’ve found the exact pitch/speed the files should be, but I’m not sure if from an archival/lossless point of view it would it be better to join all the FLACs together again before pitch-shifting and then split them back apart (a lot of work), or would it be ok just applying the pitch-shift to each FLAC file as it stands?
There is no difference.
Eg, from a really geeky technical point of view does pitch shifting involve some sort of rounding up/down interpolation which could render the beginning/end of a file different than if that section was in place within the original recording?
Technically, yes. Digital audio is quantized in both time and amplitude and there is interpolation and rounding of every sample, not just at the beginning & end. But, it shouldn’t cause any audible problems and it should be better than playing at the wrong speed. Digital audio isn’t “mathematically perfect” but it’s better than human hearing (at “CD quality” or better) and far-far better than cassette or vinyl.
You could probably pitch-shift up-and-down and back to the original pitch more than 10 times before you hear anything. That’s assuming you don’t go too far above the Nyquist limit (of half the sample rate) where you’d loose the high frequencies.
…Even a digital volume change involves rounding but It’s not considered to be a lossy process. The levels may be adjusted many times in a professional audio production and nobody gives it a 2nd thought. …And if you digitize your cassette ore than once, you’ll get different “digital data” every time. But again it won’t make an audible difference.
Changing pitch and tempo independently (which you are not doing) is more complex and is prone to audible artifacts.
Digital audio isn’t “mathematically perfect” but it’s better than human hearing (at “CD quality” or better) and far-far better than cassette or vinyl.