I know people don’t like AI stuff, but reverse engineering of an old Win16 MPEG-1 Layer 1 encoder application is not something I could make myself or have skills in, despite being a software developer for 10 years. So, here it is: another MP1/PASC encoder CLI tool, that perfectly represents what the old boy DCC2WAV does, and finally brings 24-bit support to it.
The reason why I did this is because I don’t like how other MP1/MP2 encoders work: they can’t encode full 20kHz band at 384 kbps, leaving lots of junk in treble range; they never reached this quality of mid/bass range that DCC2WAV has. I’ve posted some tests of DCC2WAV earlier: Is PASC codec that good? . I think it is the best among other MP1/MP2 tools, feel free to prove me wrong.
The open question for me is about 48 khz support, wav2pasc introduces encoding of 48k MP1 streams, but I’m not sure if these are DCC-Studio compatible or can be recorded on a cassette at all; currently wav2pasc shows warning about possible incompatibility of these files with deck recording.
Thanks, actually I didn’t even chose the language, the AI did, but I agree that C is still the most simple and universal approach in CLI utilities working with math and binary streams. Sorry, I may be quite sceptic about Rust.
My knowledge about PASC used in Philips decks is incomplete, but afaik Adaptive Mode of DCC2WAV was something very close to the original hardware algorithm, and possibly even better. More modern psychoacoustics? If this modern psychoacoustic is that we see in twolame/mp2enc, then I clearly don’t like it compared to DCC2WAV as I said. Limitations of MPEG-1 Layer I/II might be the problem, or codec developers don’t want to push too many effort into dead end technology - I don’t know. I find DCC2WAV already fine with what can ever be done with such a simple audio coding technology, but no doubt there are things can be improved. We have Musepack as an example, though it also has lotta more coding tools in it’s pockets.
If your ears are good enough to hear irregularities in other encoders/decoders, (especially in the high frequencies), can you compare your software to the NCH Switch encoder/decoder? The same company also has an audio editor called WavePad. If I remember correctly, they are free for non-commercial purposes and cheap for professional licenses (I use a commercial license to encode the DCC Museum releases). In a world where MP1 is almost completely forgotten, NCH is a rare exception. And apparently they care: When I pointed out to them that their MP1 encoder wasn’t writing the first frame to the output file, they fixed it.
Hey, yeah, I did little research on NCH already. It clearly have better quality on trebles than DCC2WAV, but it’s overall bandwidth is hard limited to 14 kHz. DCC2WAV sacrifices treble quality to some “okay” point in order to achieve full bandwidth 20+ kHz encoding, still remaining bass/mid range quality superior to others. The listening: a fellow audiophile loved it, and I trust him; my hearing is not as good.
Masking (minus over the source signal), lower is better:
Interesting! NCH must be using a more modern psycho-acoustic algorithm. I’m guessing DCC2WAV uses the algorithm described in the DCC standard of course.
