Interestingly I’ve recently found that master quality is more important than file/compression format at high quality. That is,

Rip a shitty 80s mastered CD, rip to wav, flac, etc.

Get a remastered version of the same album (watch out for loudness war, avoid 2000s), and rip 192+ mp3 or 128+ opus, high quality settings.

Compare. I’ve found the new master lossy sounds better than my old flacs, due to the better source.

Still, space is so cheap I’m mostly getting 16-44 or 48k flacs for new stuff these days, the cost of space is under ten dollars and can reencode forever.

> (watch out for loudness war, avoid 2000s)

I think the coincidence (in time, not necessarily causal direction) of master quality issues and growth in lossy codecs is probably one reason people get confused by this. People are hearing crappy recordings, listening to it on mp3s, and drawing conclusions.

I think measuring this stuff is tricky because there's so many variables involved, from the master, to the compression (or lack thereof), to the amps, speakers, quantifying perception, etc. I agree with you completely but suspect there's a bit more to it.

This is very true. Floyd Toole says in his book for a popular audience that with modern equipment, what you're hearing is roughly 1/3rd the speakers, 1/3rd the room, and 1/3rd the production quality of the recording.

Your insight on source quality is very true, and also applicable across all media, from biblical manuscripts to Japanese anime.

Yes, it isn't a new idea.

However I see geeks debating file formats ad nauseum, and I did too for many years as well when storage was expensive. But when aiming for best quality, master quality becomes the primary factor.

>You'll find that there's no audible benefit to lossless compression.*

* When the compression is done well. Shit compression makes itself known fast. If that shit compression possibly sounds okayish on 2000s style computer speakers, connecting the output to real speakers makes it even more obvious.

Not all compression is equal even if they do use the same codec.

You're conflating audio (dynamic range) compression with data compression. Dynamic range compression is an artifact of the mastering process. It very much can be done well or done poorly. Lossless data compression like FLAC isn't something that can be done "wrong". It's just encoding redundancies in the audio samples to save space. It's lossless so it will spit out the exact audio as was originally encoded.

I was writing a reply along the same lines, when I realized that we're both misreading the comment.

The original point: lossless compression is not worth it, lossy is OK.

The counter-claim: only when the lossy compression is done well.

This is not about audio compression.

Thank you, but it is about audio compression. You don't use this type of compression on text files. So, I'd agree with the other commenter in that I'm not talking about dynamic compression with hard/soft knees/attack and what not. Lossy compression done badly is horrible listening experience. It's all swishing and swooshing almost like being underwater. The top end is just all mushed together kind of like when you have a head cold and everything is all swimming sounding in your head.

There's some real confusion here.

There is:

   * dynamic range compression, referred to above as "audio compression" (confusingly)

   * psycho-acoustic aka "lossy" data compression

   * lossless data compression

The last two have nothing to do with the first.

>* dynamic range compression, referred to above as "audio compression" (confusingly)

Because when you're in a recording/editing session it is referred to as compression. As is in, "we need to use some some audio compression for that last bit". Anyone within earshot would not think someone is suggesting throwing that last bit throw an MP3 encoder.

Even so there's still the question of archival. FLAC is good for a master copy, but lossy formats mean you lose quality on every re-encode.