ayVGM - A lite version of VGM for AY-3-8910

Page 3/3
1 | 2 |

By aoineko

Paladin (808)

aoineko's picture

20-02-2022, 19:35

I updated my file format (I renamed LVGM) to include MSX-Music, MSX-Audio, SCC/SCC+ and secondary PSG data :
https://github.com/aoineko-fr/MSXgl/blob/main/engine/src/lvg...

I removed the support of the two simultaneous frequencies (50/60 Hz) because it prevents to go further in the optimizations. I'll add an option to rearrange the accesses to the registers during a frame; in particular, to make them correspond to the different packing patterns.

I haven't updated my VGM converter and my replayer yet. I'm waiting to let my thoughts mature.

I'm interested in your feedback.

By pizzapower

Master (133)

pizzapower's picture

27-11-2022, 17:27

aoineko wrote:

I updated my file format (I renamed LVGM) to include MSX-Music, MSX-Audio, SCC/SCC+ and secondary PSG data :
https://github.com/aoineko-fr/MSXgl/blob/main/engine/src/lvg...

I removed the support of the two simultaneous frequencies (50/60 Hz) because it prevents to go further in the optimizations. I'll add an option to rearrange the accesses to the registers during a frame; in particular, to make them correspond to the different packing patterns.

I haven't updated my VGM converter and my replayer yet. I'm waiting to let my thoughts mature.

I'm interested in your feedback.

I found that most of the SCC+PSG songs are still way too big to be used in something that is not a MSXDOS music player. A 1 minute song with those chips can use as much as 50kb of RAM/ROM, which is impractical, even with compression.

Have you seen this project? Maybe it can be a source of some new ideas.

By aoineko

Paladin (808)

aoineko's picture

27-11-2022, 18:27

The SCC music I tested with the VGM exporter range from 14 to 185 KB. Smile
I think I can reduce the size by about 1/3 with my lVGM tool (it is not called ayVGM anymore).
But yes, it won't be enough in most ROM games even with a 16 KB mapper.

For the moment I put the project on hold; I prefer to concentrate on adding support for a SCC tracker like Realfun 3 (in addition to Trilo Tracker already added) but if one day I have nothing better to do, I will probably look again at the issue of compressing SCC VGMs.

In my humble opinion, the most interesting approach would be to try to detect repeating patterns.
But for this to be really effective, the data from the different SCC channels would have to be de-interleaved.
It's not complicated, but I'll have to rework completely how my tool handles the data.

Page 3/3
1 | 2 |