About Lossy
We all love good music. More recently, the audio CD was good digital music. This is 44100 Hz, stereo, 16 bits (linear) per channel, not compressed in any way, which means, according to Wikipedia, 1411.2 kbps.
But at the end of the 20th century, in the era of the birth of multimedia, when music began to be played not only on players, but also on computers, it turned out that the audio CD (that is, naked PCM) is even better. . compress. There was, for example, Microsoft ADPCM, which compressed this case a bit, without losing quality, in WAV files. But generally speaking, the original 44 kHz stereo would still require a lot of space this way. Hence, the quality dropped to 22 kHz mono. One of the first multimedia albums of that time: “Immersion” from the group “Nautilus Pompilius”, is still around, and I did.
So MP3 won. To store and distribute compressed music. At 128 kbps “CD Quality”.
MP3 came up strangely. Technically, this is MPEG-1 Audio Layer 3. A layer for compressing audio data into a modern, progressive standard for storing video data on Video CDs. Just packed in its own .mp3 file format. The video CD is no longer interesting to anyone. The following MPEG-2 standard is used in DVD and digital television broadcasts (not HD). And the next MPEG-4 standard is now used for HD video and continues to evolve.
MP3 was revolutionary. It was (almost) the first lossy compression format. When we don’t try to preserve everything that was in the original signal, but, based on some psychoacoustic model, we cut out what a person is not going to hear anyway, and compress the rest. Like JPEG.
Then I tried digitizing the accumulated audio collection. Compact cassettes (just “cassettes”, but more correctly “compact cassettes”) turned out to be complete shit. The frequency range is such that it makes no sense to sample with more than 22 kHz. There were no reel-to-reel recorders in the house. But vinyl records shook the sound quality. With good equipment, you can draw better quality than a CD. You just need to get rid of the clicks.
And then I realized that MP3 is shit too. At these same 128 kbps, the sound quality suffers greatly. And the scariest thing is that vile metallic hues appear where they shouldn’t be. My ears need at least 192 kbps, and the more the better.
Let’s take a hint from a famous punk rock band in the past. Like FLAC. It is such a modern lossless compression standard that it has successfully replaced WAV. Because it is free.
The original is CD quality, so frequencies up to 22 kHz are present as expected.
Original flac
We are going to harvest with FFmpeg, or rather with LAME.
At 320 kbps and 256 kbps, the spectrogram looks almost like the original.
At 192 kbps, there are signs of a 16 kHz cutoff. The spectrogram “darkens”, apparently, the psychoacoustic model has cut something out. By ear, the higher frequency “bursts” really disappeared.
MP3 192 kbps
At the notorious 128 kbit / s, everything is already specifically cut off at 16 kHz. Background sounds are “fuzzy” and begin to bubble. Nothing to do with the original in terms of enjoying the musical details.
MP3 128 kbps
But you can do 64 kbps in MP3. The stereo is gone. Everything gurgles terribly and irritates with completely strange sounds.
