Merge pull request #714 from heygen-com/fix/713-webaudio-playback-rate

fix(core): thread playback rate into WebAudio audio sources
This commit is contained in:
James Russo
2026-05-11 10:40:56 -04:00
committed by GitHub
3 changed files with 204 additions and 43 deletions
+2
View File
@@ -1589,6 +1589,7 @@ export function initSandboxRuntimeModular(): void {
onSetPlaybackRate: (rate) => {
applyPlaybackRate(rate);
if (state.transportClock) state.transportClock.setRate(state.playbackRate);
webAudio.setRate(state.playbackRate);
},
onEnablePickMode: () => picker.enablePickMode(),
onDisablePickMode: () => picker.disablePickMode(),
@@ -1858,6 +1859,7 @@ export function initSandboxRuntimeModular(): void {
clock.now(),
vol * state.bridgeVolume,
gen,
state.playbackRate,
);
});
}
@@ -1,6 +1,49 @@
import { describe, it, expect, vi } from "vitest";
import { WebAudioTransport } from "./webAudioTransport";
function createMockAudioContext(currentTime = 100) {
const startFn = vi.fn();
const sourceNode = {
buffer: null as AudioBuffer | null,
playbackRate: { value: 1 },
start: startFn,
stop: vi.fn(),
disconnect: vi.fn(),
connect: vi.fn(),
};
const gainNode = {
gain: { value: 1 },
connect: vi.fn(),
disconnect: vi.fn(),
};
const masterGain = {
gain: { value: 1 },
connect: vi.fn(),
};
const ctx = {
currentTime,
state: "running",
resume: vi.fn(),
createBufferSource: vi.fn(() => sourceNode),
createGain: vi.fn(() => gainNode),
destination: {},
close: vi.fn(),
};
return { ctx, sourceNode, gainNode, masterGain, startFn };
}
function setupTransport(currentTime = 100) {
const transport = new WebAudioTransport();
const mock = createMockAudioContext(currentTime);
(transport as unknown as { _ctx: unknown })._ctx = mock.ctx;
(transport as unknown as { _masterGain: unknown })._masterGain = mock.masterGain;
const gen = transport.startGeneration();
return { transport, mock, gen };
}
const mockBuffer = {} as AudioBuffer;
const mockEl = { muted: false } as HTMLMediaElement;
describe("WebAudioTransport", () => {
it("tracks play generation for async race prevention", () => {
const transport = new WebAudioTransport();
@@ -82,45 +125,6 @@ describe("WebAudioTransport", () => {
});
describe("schedulePlayback timing", () => {
function createMockAudioContext(currentTime = 100) {
const startFn = vi.fn();
const sourceNode = {
buffer: null as AudioBuffer | null,
start: startFn,
connect: vi.fn(),
};
const gainNode = {
gain: { value: 1 },
connect: vi.fn(),
};
const masterGain = {
gain: { value: 1 },
connect: vi.fn(),
};
const ctx = {
currentTime,
state: "running",
resume: vi.fn(),
createBufferSource: vi.fn(() => sourceNode),
createGain: vi.fn(() => gainNode),
destination: {},
close: vi.fn(),
};
return { ctx, sourceNode, gainNode, masterGain, startFn };
}
function setupTransport(currentTime = 100) {
const transport = new WebAudioTransport();
const mock = createMockAudioContext(currentTime);
(transport as unknown as { _ctx: unknown })._ctx = mock.ctx;
(transport as unknown as { _masterGain: unknown })._masterGain = mock.masterGain;
const gen = transport.startGeneration();
return { transport, mock, gen };
}
const mockBuffer = {} as AudioBuffer;
const mockEl = { muted: false } as HTMLMediaElement;
it("starts in-progress clips immediately with correct buffer offset", async () => {
const { transport, mock, gen } = setupTransport(100);
@@ -161,4 +165,120 @@ describe("WebAudioTransport", () => {
expect(mock.startFn).toHaveBeenCalledWith(0, 0);
});
});
describe("playback rate", () => {
it("sets sourceNode.playbackRate.value when rate is provided", async () => {
const { transport, mock, gen } = setupTransport(100);
await transport.schedulePlayback(mockEl, mockBuffer, 5, 0, 8, 1, gen, 2);
expect(mock.sourceNode.playbackRate.value).toBe(2);
});
it("defaults rate to 1 when not provided", async () => {
const { transport, mock, gen } = setupTransport(100);
await transport.schedulePlayback(mockEl, mockBuffer, 5, 0, 8, 1, gen);
expect(mock.sourceNode.playbackRate.value).toBe(1);
});
it("scales delay by rate for future clips so they fire at the right wallclock", async () => {
const { transport, mock, gen } = setupTransport(100);
// compStart=10, compositionTime=2, rate=2 → 8s of comp time = 4s wallclock
await transport.schedulePlayback(mockEl, mockBuffer, 10, 0, 2, 1, gen, 2);
expect(mock.startFn).toHaveBeenCalledWith(104, 0);
});
it("keeps in-progress buffer offset at elapsed + mediaStart regardless of rate", async () => {
const { transport, mock, gen } = setupTransport(100);
await transport.schedulePlayback(mockEl, mockBuffer, 5, 0, 8, 1, gen, 2);
expect(mock.startFn).toHaveBeenCalledWith(0, 3);
});
it("setRate updates active sources in place", async () => {
const { transport, mock, gen } = setupTransport(100);
await transport.schedulePlayback(mockEl, mockBuffer, 5, 0, 8, 1, gen, 1);
expect(mock.sourceNode.playbackRate.value).toBe(1);
transport.setRate(2);
expect(mock.sourceNode.playbackRate.value).toBe(2);
});
it("setRate before any sources are scheduled does not throw", () => {
const transport = new WebAudioTransport();
expect(() => transport.setRate(2)).not.toThrow();
});
it("setRate is a no-op when the rate is unchanged", async () => {
const { transport, mock, gen } = setupTransport(100);
await transport.schedulePlayback(mockEl, mockBuffer, 5, 0, 8, 1, gen, 2);
mock.ctx.currentTime = 100.5;
const timeBefore = transport.getTime();
transport.setRate(2);
const timeAfter = transport.getTime();
expect(timeAfter).toBe(timeBefore);
// No re-anchor, so the next 0.5s of wallclock still maps to 1s of comp time.
mock.ctx.currentTime = 101;
expect(transport.getTime()).toBeCloseTo(10, 10);
});
it("setRate clamps non-finite or non-positive values to 1", async () => {
const { transport, mock, gen } = setupTransport(100);
await transport.schedulePlayback(mockEl, mockBuffer, 5, 0, 8, 1, gen, 2);
expect(mock.sourceNode.playbackRate.value).toBe(2);
transport.setRate(Number.NaN);
expect(mock.sourceNode.playbackRate.value).toBe(1);
transport.setRate(2);
transport.setRate(0);
expect(mock.sourceNode.playbackRate.value).toBe(1);
transport.setRate(2);
transport.setRate(-1);
expect(mock.sourceNode.playbackRate.value).toBe(1);
});
it("getTime advances at the configured rate", async () => {
const { transport, mock, gen } = setupTransport(100);
await transport.schedulePlayback(mockEl, mockBuffer, 5, 0, 8, 1, gen, 2);
// At schedule time, ctx.currentTime=100, compositionTime=8.
expect(transport.getTime()).toBeCloseTo(8, 10);
// Advance the audio-context clock by 0.5 wallclock seconds; at rate=2,
// composition time should have advanced 1s.
mock.ctx.currentTime = 100.5;
expect(transport.getTime()).toBeCloseTo(9, 10);
});
it("getTime tracks composition time after a mid-playback setRate", async () => {
const { transport, mock, gen } = setupTransport(100);
await transport.schedulePlayback(mockEl, mockBuffer, 5, 0, 8, 1, gen, 1);
expect(transport.getTime()).toBeCloseTo(8, 10);
// 0.5s passes at rate=1 → composition time = 8.5
mock.ctx.currentTime = 100.5;
expect(transport.getTime()).toBeCloseTo(8.5, 10);
// Bump rate to 2 — composition time should NOT jump.
transport.setRate(2);
expect(transport.getTime()).toBeCloseTo(8.5, 10);
// Another 0.5s wallclock at rate=2 → composition time = 9.5
mock.ctx.currentTime = 101;
expect(transport.getTime()).toBeCloseTo(9.5, 10);
});
});
});
+43 -4
View File
@@ -1,5 +1,10 @@
import { swallow } from "./diagnostics";
function normalizeRate(rate: number): number {
if (!Number.isFinite(rate) || rate <= 0) return 1;
return rate;
}
export type ScheduledSource = {
el: HTMLMediaElement;
sourceNode: AudioBufferSourceNode;
@@ -15,7 +20,12 @@ export class WebAudioTransport {
private _bufferCache = new Map<string, AudioBuffer>();
private _activeSources: ScheduledSource[] = [];
private _masterGain: GainNode | null = null;
private _scheduleOffset = 0;
// Composition-time reference frame: at AudioContext time `_rateAnchorCtx`,
// composition time was `_rateAnchorComp`, and time has been advancing at
// `_rate` composition-seconds per wallclock-second since.
private _rateAnchorCtx = 0;
private _rateAnchorComp = 0;
private _rate = 1;
private _paused = true;
private _playGeneration = 0;
@@ -36,7 +46,7 @@ export class WebAudioTransport {
getTime(): number {
if (!this._ctx || this._paused) return -1;
return this._ctx.currentTime - this._scheduleOffset;
return this._rateAnchorComp + (this._ctx.currentTime - this._rateAnchorCtx) * this._rate;
}
async decodeAudioElement(el: HTMLMediaElement): Promise<AudioBuffer | null> {
@@ -73,6 +83,7 @@ export class WebAudioTransport {
compositionTime: number,
volume: number,
generation: number,
rate = 1,
): Promise<ScheduledSource | null> {
if (!this._ctx || !this._masterGain) return null;
if (generation !== this._playGeneration) return null;
@@ -83,8 +94,11 @@ export class WebAudioTransport {
}
if (generation !== this._playGeneration) return null;
const safeRate = normalizeRate(rate);
const sourceNode = this._ctx.createBufferSource();
sourceNode.buffer = buffer;
sourceNode.playbackRate.value = safeRate;
const gainNode = this._ctx.createGain();
gainNode.gain.value = volume;
@@ -93,12 +107,14 @@ export class WebAudioTransport {
const elapsed = compositionTime - compositionStart;
const scheduledAt = this._ctx.currentTime;
this._scheduleOffset = scheduledAt - compositionTime;
this._rate = safeRate;
this._rateAnchorCtx = scheduledAt;
this._rateAnchorComp = compositionTime;
if (elapsed >= 0) {
sourceNode.start(0, elapsed + mediaStart);
} else {
const delay = -elapsed;
const delay = -elapsed / safeRate;
sourceNode.start(scheduledAt + delay, mediaStart);
}
@@ -123,6 +139,29 @@ export class WebAudioTransport {
}
}
/**
* Rebases the composition-time reference frame before swapping rate so
* `getTime()` stays continuous across the change. Sources scheduled to
* start in the future keep their original wallclock start time — callers
* that need rate-correct future starts should `stopAll()` and reschedule.
*/
setRate(rate: number): void {
const safeRate = normalizeRate(rate);
if (safeRate === this._rate) return;
if (this._ctx && !this._paused) {
this._rateAnchorComp = this.getTime();
this._rateAnchorCtx = this._ctx.currentTime;
}
this._rate = safeRate;
for (const source of this._activeSources) {
try {
source.sourceNode.playbackRate.value = safeRate;
} catch (err) {
swallow("webAudioTransport.setRate", err);
}
}
}
stopAll(): void {
for (const source of this._activeSources) {
try {