curioustorvald/tsvm
1
0
Fork 0
mirror of https://github.com/curioustorvald/tsvm.git synced 2026-06-06 05:28:31 +09:00
Code Issues Packages Projects Releases Wiki Activity

audio changes

Browse Source
This commit is contained in:
minjaesong
2026-04-16 21:58:06 +09:00
parent 6aa2542bb8
commit 7d899936e2
4 changed files with 31 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files

39
assets/disk0/tvdos/include/psg.mjs
View File

@@ -119,7 +119,7 @@ function mixInto(buf, lengthSec, offsetSec, op, amp, pan, sampleFn) {
// ── Waveform generators ─────────────────────────────────────────────────────
function makeSquare(buf, length, offset, freq, duty, op, amp, pan) {
function makeSquare(buf, length, offset, freq, duty, op, amp, pan, phaseOffset) {
// buffer: [Uint8Array, Uint8Array] or native buffer
// length: in seconds
// offset: in seconds
@@ -128,13 +128,16 @@ function makeSquare(buf, length, offset, freq, duty, op, amp, pan) {
// op: add / mul / sub; default: add
// amp: 0.0 to 1.0; default: 0.5
// pan: -1.0 to 1.0; default: 0.0
// phaseOffset: optional absolute-time base (seconds) added to phase calc only,
// not to the buffer write position — use to ensure phase continuity
// across successive calls (e.g. frame boundaries).
if (duty == null) duty = 0.5
if (op == null) op = 'add'
if (amp == null) amp = 0.5
if (pan == null) pan = 0.0
const tBase = (phaseOffset || 0) + offset
mixInto(buf, length, offset, op, amp, pan, function(i) {
const t = offset + i / HW_SAMPLING_RATE
const phase = (t * freq) % 1
const phase = ((tBase + i / HW_SAMPLING_RATE) * freq) % 1
return (phase < duty) ? 1.0 : -1.0
})
}
@@ -225,7 +228,7 @@ function lfsrAdvance(state, steps, mode) {
const LFSR_PERIOD_LONG = 32767 // mode 0
const LFSR_PERIOD_SHORT = 93 // mode 1
function makeNoise(buf, length, offset, freq, type, op, amp, pan) {
function makeNoise(buf, length, offset, freq, type, op, amp, pan, phaseOffset) {
// buffer: [Uint8Array, Uint8Array] or native buffer
// length: in seconds
// offset: in seconds
@@ -238,20 +241,23 @@ function makeNoise(buf, length, offset, freq, type, op, amp, pan) {
// op: add / mul / sub; default: add
// amp: 0.0 to 1.0; default: 0.5
// pan: -1.0 to 1.0; default: 0.0
// phaseOffset: optional absolute-time base (seconds) added to phase/LFSR calc only —
// see makeSquare for details.
//
// LFSR types (1 and 2) are deterministic given (offset, freq): calling with
// monotonically advancing offset values produces a seamless noise stream
// LFSR types (1 and 2) are deterministic given (phaseOffset+offset, freq): calling
// with monotonically advancing phaseOffset+offset produces a seamless noise stream
// across frames. White noise types (-1, 0) are random per call.
if (op == null) op = 'add'
if (amp == null) amp = 0.5
if (pan == null) pan = 0.0
const tBase = (phaseOffset || 0) + offset
if (type === -1) {
// 8-bit white: new random float in [-1, 1] each clock period
let prevClock = -1
let noiseVal = 0.0
mixInto(buf, length, offset, op, amp, pan, function(i) {
const currentClock = Math.floor((offset + i / HW_SAMPLING_RATE) * freq) | 0
const currentClock = Math.floor((tBase + i / HW_SAMPLING_RATE) * freq) | 0
if (currentClock !== prevClock) {
prevClock = currentClock
noiseVal = Math.random() * 2.0 - 1.0
@@ -263,7 +269,7 @@ function makeNoise(buf, length, offset, freq, type, op, amp, pan) {
let prevClock = -1
let noiseVal = 1.0
mixInto(buf, length, offset, op, amp, pan, function(i) {
const currentClock = Math.floor((offset + i / HW_SAMPLING_RATE) * freq) | 0
const currentClock = Math.floor((tBase + i / HW_SAMPLING_RATE) * freq) | 0
if (currentClock !== prevClock) {
prevClock = currentClock
noiseVal = (Math.random() >= 0.5) ? 1.0 : -1.0
@@ -274,13 +280,13 @@ function makeNoise(buf, length, offset, freq, type, op, amp, pan) {
// LFSR-based noise (types 1 and 2)
const mode = (type === 2) ? 1 : 0
const period = (mode === 0) ? LFSR_PERIOD_LONG : LFSR_PERIOD_SHORT
// Advance to deterministic position for this offset so consecutive frame
// calls with advancing offsets produce a seamless noise stream.
const startClock = Math.floor(offset * freq) | 0
// Advance to deterministic position for this tBase so consecutive frame
// calls with monotonically advancing phaseOffset produce a seamless noise stream.
const startClock = Math.floor(tBase * freq) | 0
let lfsr = lfsrAdvance(1, startClock % period, mode)
let prevClock = startClock
mixInto(buf, length, offset, op, amp, pan, function(i) {
const currentClock = Math.floor((offset + i / HW_SAMPLING_RATE) * freq) | 0
const currentClock = Math.floor((tBase + i / HW_SAMPLING_RATE) * freq) | 0
const delta = currentClock - prevClock
if (delta > 0) {
const steps = delta % period
@@ -292,17 +298,18 @@ function makeNoise(buf, length, offset, freq, type, op, amp, pan) {
}
}
function makeAliasedTriangleNES(buf, length, offset, freq, duty, op, amp, pan) {
function makeAliasedTriangleNES(buf, length, offset, freq, duty, op, amp, pan, phaseOffset) {
// NES APU triangle — quantised to the authentic 32-step, 4-bit (0..15) staircase.
// The 32-step sequence is: 15,14,...,1,0, 0,1,...,14,15 (descending then ascending).
// This mirrors the real NES triangle DAC which has 32 equal-height steps per period.
// duty parameter is accepted for API symmetry but ignored (NES triangle is always symmetric).
// phaseOffset: optional absolute-time base (seconds) — see makeSquare for details.
if (op == null) op = 'add'
if (amp == null) amp = 0.5
if (pan == null) pan = 0.0
const tBase = (phaseOffset || 0) + offset
mixInto(buf, length, offset, op, amp, pan, function(i) {
const t = offset + i / HW_SAMPLING_RATE
const phase = (t * freq) % 1
const phase = ((tBase + i / HW_SAMPLING_RATE) * freq) % 1
const step32 = Math.floor(phase * 32) | 0 // 0..31
// step 0..15: descend from 15 to 0; step 16..31: ascend from 0 to 15
const level = (step32 < 16) ? (15 - step32) : (step32 - 16)
@@ -341,7 +348,7 @@ function sendBuffer(buf, playhead, offsetSec, lengthSec, stagingPtr) {
}
// Wait for room in the playback queue (mirrors playwav.js idiom)
// while (audio.getPosition(playhead) > 2) sys.sleep(2)
audio.putPcmDataByPtr(stagingPtr, take * 2, playhead)
audio.putPcmDataByPtr(playhead, stagingPtr, take * 2, 0)
audio.setSampleUploadLength(playhead, take * 2)
audio.startSampleUpload(playhead)
remaining -= take

5
terranmon.txt
View File

@@ -1978,6 +1978,11 @@ Sound Adapter
Endianness: little
TSVM Sound Adapter is consisted of 4 playheads, each playhead is capable of playing one PCM or Tracker track.
Synchronisation between playheads are not guaranteed. Do not play music in multiple tracks.
Memory Space
0..114687 RW: Sample bin

2
tsvm_core/src/net/torvald/tsvm/AudioJSR223Delegate.kt
View File

@@ -10,7 +10,7 @@ import net.torvald.tsvm.peripheral.MP2Env
*
* NOTES:
* 1. tracker mode is currently unimplemented.
* 2. PCM upload buffer (accessed by `putPcmDataByPtr`) is shared between four playheads
* 2. Synchronisation between playheads are not guaranteed. Do not play music in multiple tracks.
*
* ## How to upload PCM audio into a playhead
*

4
tsvm_core/src/net/torvald/tsvm/peripheral/AudioAdapter.kt
View File

@@ -41,7 +41,7 @@ private class RenderRunnable(val playhead: AudioAdapter.Playhead) : Runnable {
// printdbg("P${playhead.index+1} go back to spinning")
Thread.sleep(12)
Thread.sleep(6)
}
else if (playhead.isPlaying && writeQueue.isEmpty) {
printdbg("!! QUEUE EXHAUSTED !! QUEUE EXHAUSTED !! QUEUE EXHAUSTED !! QUEUE EXHAUSTED !! QUEUE EXHAUSTED !! QUEUE EXHAUSTED ")
@@ -49,7 +49,7 @@ private class RenderRunnable(val playhead: AudioAdapter.Playhead) : Runnable {
// TODO: wait for 1-2 seconds then finally stop the device
// playhead.audioDevice.stop()
Thread.sleep(12)
Thread.sleep(6)
}
}