nullsound: fix YM2610 reset before stream playback
FM channels can't be stopped immediately, as their operators can
only be configured to release their note.
Implement the reset by configuring the fastest release possible
and stopping the operator in sequence. This is sufficient to
remove any unpleasant "pop" sound that was appearing while
stopping a NSS stream and playing another one immediately.
Also update the snd_stream_stop to call ym2610_reset so that
all channels are effectively stopped when this function is called.
Currently ADPCM-A and FM opcodes (instrument, note on...) are
two bytes long and they include the action's target channel.
Optimize the space used by defining a 10 new CTX opcodes.
A CTX is a 1 byte opcode that defines the scope of the next
ADPCM-A or FM action.
ADPCM-A and FM have their own context of action. Each time
a ADPCM-A or FM opcode is evaluated, the according context
is incremented by 1, so that the scope of the next opcode
will automatically be the next ADPCM-A or FM channel.
This approach allows efficient NSS stream storage in ROM, and
will be reused when implementing the next NSS opcodes
like e.g. volume, detune...
The original ADPCM-A and FM opcodes are now deprecated and
will be replaced eventually.
vromtool.py automatically packs ADPCM samples into one or more
VROMs, and generates ASM defines to export the samples positions
in VROM to the z80 sound driver.
furtool.py extracts the FM and ADPCM instruments define in a
of a Furnace module, and generate instrument data ready to be
consumed by nullsound NSS stream player. It also extract the
ADPCM samples from the Furnace module in a sample map that
can be consumed by vromtool.py to generate VROMs.
adpcmtool: fix incorrect padding of ADPCM-A samples
When a WAV file is converted to ADPCM, adpcmtool automatically
pads it to a multiple of 512 samples so that the generated
ADPCM sample be a multiple of 256 bytes.
The padding is incorrect for ADPCM-A (it works for ADPCM-B), so
fix it accordingly.
This is a initial commit to play a music encoded in the nullsound stream
format (NSS), which is as a series of opcodes that can configure various
parts of the YM2610, wait for timers, and control music loop.
The NSS format is still considered experimental and will change in
future commits.
