Everything you hear here comes from running the cartridge's own sound engine (PASO_DE_CANAL, 0x4111, once per channel from the interrupt, and the tracks each routine points the channel records at, 0xE020, 0xE028 and 0xE030) on an emulated Z80, frame by frame at 60 per second, just as the game calls it. What the engine writes to the PSG is turned into sound with an AY-3-8910 (tools/musica.py). Each MP3 is the whole piece up to where it starts again, plus 8 seconds of the repeat fading out; those that do not repeat end where the engine ends.
The PSG writes have been compared frame by frame with those of openMSX (C-BIOS MSX1 japonés). In every frame the PSG ends up the same:
The timbre is not checked: it is a textbook AY-3-8910, not a recording of the emulator. The probes are tools/omsx_psg.tcl and tools/coteja_psg.py.
This game has no routine that takes a sound number: whoever wants something to play writes the pointer to its track into a channel record, and PASO_DE_CANAL reads it note by note. Each record is eight bytes: pointer, what is left of the note, the channel's PSG register (0, 2 or 4), the volume... It is the same player as in other Konami cartridges of the time. The openMSX check was done by writing those same records during play and comparing what the PSG receives from then on. The engine does not write the PSG mixer except for noise: start-up leaves it set, and here it starts from the same value, 0xB8.
The start-up tune (0x01)
Asked for by Y_LA_FASE_DE_LOS_DOS (0x442B): tres canales, programas de 0x5D97, 0x5DE9 y 0x5E41; lasts 6.0 s.
The end of the game (0x02)
Asked for by FIN_DE_PARTIDA (0x4596): dos canales, programas de 0x5FB6 y 0x5FD8; lasts 2.7 s.
Stage cleared (0x03)
Asked for by la rutina de 0x5411: dos canales, programas de 0x5F18 y 0x5F5E; lasts 5.4 s.