Sound
The WonderSwan features the following sound hardware:
- Four audio channels:
- channel 1 - wavetable (32 x 4-bit samples),
- channel 2 - wavetable or 8-bit unsigned PCM sample,
- channel 3 - wavetable with optional hardware sweep,
- channel 4 - wavetable or LFSR noise,
- Hyper Voice(color) - headphone output exclusive 16-bit stereo PCM output,
- 24000 Hz output:
- internal speaker - 8-bit, mono,
- headphone output - 16-bit, stereo.
The sound is mixed as follows:
Ch1 Ch2 Ch3 Ch4 Glossary: L = left, R = right, M = mono
| | | | Mnn = nn-bit mono
8 8 8 8 port $98 port $nn = value port
| | | | /
_V___V___V___V_ / _____ _____
| |--L10------+------| | | |
| + | | | + |--M11-->| >>r |--M8--> Speaker output
|_______________|--R10------|-+----|_____| / |_____|
/ _|_|_ / (r=0...3)
/ | | /
port $96 | <<5 | port $9A
|_____|
| |
| |
port $68 port $64 | |
/ _______________ \ __V_V__ _____
L8-->| (Color only) |--L16-->| |----------L16---| |
| Hyper Voice | | + + | | I2S |------> Headphone output
R8-->|_______________|--R16-->|_______|----------R16---|_____|
\ /
port $69 port $66
Hyper Voice functionality is documented on its own sub-page.
I/O Ports
Sound Channel Frequency ($80, $81; $82, $83; $84, $85; $86, $87)
15 bit 8 7 bit 0
---- ---- ---- ----
.... .ddd dddd dddd
||| |||| ||||
+++--++++-++++- Frequency divisor
This frequency is used for wavetable output. It is calculated as follows: sample rate = 3072000 Hz / (2048 - divisor).
Note that this refers to the sample rate of the wavetable, and needs to be scaled accordingly for a given waveform when performing music playback.
Sound Channel Volume ($88; $89; $8A; $8B)
7 bit 0 --------- llll rrrr |||| |||| |||| ++++- Right channel volume (0-15) ++++------ Left channel volume (0-15)
This volume is used for wavetable output.
The calculation to get an unsigned 8-bit sample out of a 4-bit wavetable sample / 4-bit volume pair is simple: out_sample = sample * volume. This means that the maximum sample in wavetable output mode is 15 * 15 = 225.
Sound Channel 2 Voice Sample ($89)
7 bit 0 ---- ---- ssss ssss |||| |||| ++++-++++- Unsigned 8-bit PCM sample (0-255)
This sample is used for voice output.
Sound Channel 3 Sweep Amount ($8C)
7 bit 0 ---- ---- vvvv vvvv |||| |||| ++++-++++- Value; 8-bit, signed.
This port is used for wavetable output with sweep enabled.
The signed value in this port will be added to the frequency divider for channel 3 every sweep step (as determined by the Sweep Ticks port). Wraparound is present - adding 1 to a frequency divider value of 2047 will cause it to roll over back to 0.
Sound Channel 3 Sweep Ticks ($8D)
7 bit 0 ---- ---- ...t tttt | |||| +-++++- Ticks per step - 1
This port is used for wavetable output with sweep enabled.
Every ttttt + 1 ticks, clocked at 375 Hz, the value in the Sweep Amount port will be added to the frequency divider for channel 3.
Sound Channel 4 Noise Control ($8F)
7 bit 0 ---- ---- ...e rttt | |||| | |+++- LFSR tap mode | +---- LFSR reset: 1 = reset shift register +------ LFSR enabled: 0 = off, 1 = on
This port is used for noise output.
| Tap mode | Tap bit | Sequence length |
|---|---|---|
| 0 | 14 | 32767 |
| 1 | 10 | 1953 |
| 2 | 13 | 254 |
| 3 | 4 | 217 |
| 4 | 8 | 73 |
| 5 | 6 | 63 |
| 6 | 9 | 42 |
| 7 | 11 | 28 |
Noise output works using the following algorithm, performed once per sample generated:
- Create a new bit by XORing bit 7 with the tap bit.
- Shift the LFSR register one bit to the left.
- Write the new bit as bit 0.
- Use the new bit as if it were a wavetable sample: 0 = 0, 1 = 15.
Sound Wavetable Address ($8F)
7 bit 0 ---- ---- wwww wwww |||| |||| ++++-++++- Wavetable address (bits 6-13)
The wavetable is 4 x 16 bytes long; each 16-byte block contains 32 4-bit samples for the relevant channel:
Address $00 $01 $02 $03 $04 $05 $06 $07 $08 $09 $0A $0B $0C $0D $0E $0F
98 CB ED FF FF EF BD 8A 57 24 01 00 00 21 43 76
Sample
15 .. .. .. ## ## #. .. .. .. .. .. .. .. .. .. ..
14 .. .. .# || || |# .. .. .. .. .. .. .. .. .. ..
13 .. .. #| || || || #. .. .. .. .. .. .. .. .. ..
12 .. .# || || || || |. .. .. .. .. .. .. .. .. ..
11 .. #| || || || || |# .. .. .. .. .. .. .. .. ..
10 .. || || || || || || #. .. .. .. .. .. .. .. ..
9 .# || || || || || || |. .. .. .. .. .. .. .. ..
8 #| || || || || || || |# .. .. .. .. .. .. .. ..
7 || || || || || || || || #. .. .. .. .. .. .. .#
6 || || || || || || || || |. .. .. .. .. .. .. #|
5 || || || || || || || || |# .. .. .. .. .. .. ||
4 || || || || || || || || || #. .. .. .. .. .# ||
3 || || || || || || || || || |. .. .. .. .. #| ||
2 || || || || || || || || || |# .. .. .. .# || ||
1 || || || || || || || || || || #. .. .. #| || ||
0 || || || || || || || || || || |# ## ## || || ||
Note that the higher bits - the first hexadecimal number - specify the later samples; that is to say, the nybbles of each sample are swapped relative to the wave drawing.
Sound Channel Control ($90)
7 bit 0 ---- ---- nsv. 4321 ||| |||| ||| |||+- Channel 1 enable ||| ||+-- Channel 2 enable ||| |+--- Channel 3 enable ||| +---- Channel 4 enable ||+------- Channel 2 mode: 0 = wavetable, 1 = voice |+-------- Channel 3 sweep: 0 = disable, 1 = enable +--------- Channel 4 mode: 0 = wavetable, 1 = noise
Sound Output Control ($91)
7 bit 0 ---- ---- H... hrrs | |||| | |||+- Internal speaker output enable | |++-- Internal speaker shift right value | +---- Headphone output enable +--------- Headphones connected: 1 if true
It is a good idea to set the internal speaker shift right value correctly. If the value is too low, multi-channel music will clip; if the value is too high, single-channel PCM sample playback will be very quiet.
Sound Channel 4 LFSR Register ($92, $93 read)
15 bit 8 7 bit 0 ---- ---- ---- ---- .rrr rrrr rrrr rrrr ||| |||| |||| |||| +++-++++--++++-++++- Shift register value
Sound Channel 2 Voice Volume ($94)
7 bit 0
---- ----
.... lLrR
||||
|||+- Right channel 100% volume
||+-- Right channel 50% volume
|+--- Left channel 100% volume
+---- Left channel 50% volume
This port is used for voice output.
The 50% volume have no effect if their respective 100% volume bits are set.
