Lossy audio compression algorithms work based on psychoacoustic effects. The average human ear will not detect all the “parts” in a lossless signal - there are things you can drop from the signal because:
- Human ears are most sensitive around the frequency of human speech, but less at others
- If there is a loud signal, a much more silent one very close will be masked if it occurs within a couple of milliseconds around the loud one
- There are other more subtle aspects of the human ear you can use to detect signals we just won’t notice.
So in order to determine exactly which parts of an audio signal could be dropped because we don’t hear them anyway, they measured a couple of thousand people’s listening profiles.
And they used that “average human profile” to create their algorithm.
This, of course, has a consequence which most people, including you apparently, do not understand:
The better your personal “ear” matches the average psychoacoustic model used by lossy algorithms, the better the signal will sound to you.
In other words, older people, or people with certain deficiencies in their hearing capabilities, will need higher bitrates not to notice the difference. In the 90s, I used to be happy with 192 kbps CBR MP3. But now, being an old fuck, boy, can I hear the difference.
Ironically, I can detect the difference not because my ear is trained or better, I can detect it because my ear is worse than yours!
So the whole bottom line is this: While it may be true that you, personally, do not require lossless to enjoy music to the fullest, other people do. Claiming that lossless isn’t needed by 99.9% of the population is horseshit and only demonstrates that you have no clue about how lossy compression works in the first place.
sefra1@lemmy.zip 1 day ago
The fuss is that every time you transcode to a new format you accumulatively lose quality.
So for example if you have an 320kbps mp3, but then that takes too much space so you transcode it to 192 mp3, but then you discover the opus codec is more efficient so you transcode it again, but then you want to make a fan video of the same song, so your video player transcoded it again into video friendly aac.
The quality on your final video is going contain the faults of all the files upstream.
Meanwhile if you edit the video from a lossless source, it will only get encoded once.
So it doesn’t matter for streaming, but it matters if you want to download and convert to other formats.
Substance_P@lemmy.world 1 day ago
This is a great point, currently I have tens of thousands of mp3’s that I wish I could somehow, impossibly upscale to a better codec, but those rare tracks I have in the low VBR mp3 range will never be revived.