feat(voice-input): select microphone with menu-only input levels

This commit is contained in:
yudshj authored and Yudong Han committed 2026-10-08 10:48:48 +08:00
1 parent b4443d4a90
commit 2d0360870c
18 files changed
+481 -26

No files matched your search

+37 -1
View File
@@ -62,7 +62,14 @@ it.skipIf(webSnapshotMode() === 'record')('guides voice setup, records from stan
resources.browser = browser
const page = await newEnglishPage(browser), tripwire = watchConsole(page)
await page.addInitScript(() => {
const holder = window as Window & { voiceTestRecorder?: MediaRecorder }
const holder = window as Window & { voiceTestRecorder?: MediaRecorder; voiceTestStreams?: MediaStream[] }
holder.voiceTestStreams = []
const getUserMedia = navigator.mediaDevices.getUserMedia.bind(navigator.mediaDevices)
navigator.mediaDevices.getUserMedia = async (constraints) => {
const stream = await getUserMedia(constraints)
holder.voiceTestStreams!.push(stream)
return stream
}
const NativeRecorder = window.MediaRecorder
window.MediaRecorder = class extends NativeRecorder {
constructor(stream: MediaStream, options?: MediaRecorderOptions) { super(stream, options); holder.voiceTestRecorder = this }
@@ -120,6 +127,33 @@ it.skipIf(webSnapshotMode() === 'record')('guides voice setup, records from stan
expect(recognize).not.toHaveBeenCalled()
preparation = { phase: 'standby' }
for (const listener of readinessListeners) listener()
const devicePicker = page.getByRole('button', { name: 'Input device', exact: true })
await devicePicker.click()
const level = page.getByRole('img', { name: 'Microphone input level', exact: true })
await level.waitFor()
const devices = await page.evaluate(async () => (await navigator.mediaDevices.enumerateDevices())
.filter(device => device.kind === 'audioinput' && device.deviceId !== 'default'))
const selectedDevice = devices[0]!
await page.getByRole('menuitem', { name: selectedDevice.label, exact: true }).click()
await page.getByRole('menuitem', { name: `${selectedDevice.label} Microphone input level`, exact: true }).waitFor()
await expect.poll(() => page.evaluate(() => {
const streams = (window as Window & { voiceTestStreams?: MediaStream[] }).voiceTestStreams!
return streams.filter(stream => stream.getTracks().some(track => track.readyState === 'live')).length
})).toBe(1)
await compareOrRefreshGolden(fileURLToPath(new URL('../../../snapshots/web/voice-input/device.expected.md', import.meta.url)),
await captureStableAria(page, '[role="menu"]', scaffold.workspaceCwd), webSnapshotMode())
for (const width of [1280, 360]) {
await page.setViewportSize({ width, height: 900 })
await expect.poll(async () => {
const bounds = await page.getByRole('menu').boundingBox()
return bounds !== null && bounds.x >= 0 && bounds.x + bounds.width <= width
}).toBe(true)
}
await page.keyboard.press('Escape')
await level.waitFor({ state: 'hidden' })
await expect.poll(() => page.evaluate(() => (window as Window & { voiceTestStreams?: MediaStream[] }).voiceTestStreams!
.every(stream => stream.getTracks().every(track => track.readyState === 'ended')))).toBe(true)
await page.setViewportSize({ width: 1280, height: 900 })
await page.getByRole('button', { name: 'New session', exact: true }).filter({ hasText: 'New Session' }).click()
await mic.waitFor()
await page.getByRole('button', { name: 'Plugins', exact: true }).click()
@@ -134,6 +168,8 @@ it.skipIf(webSnapshotMode() === 'record')('guides voice setup, records from stan
await input.fill(prefix)
await mic.click()
await page.getByRole('button', { name: 'Stop and transcribe', exact: true }).waitFor()
expect(await page.evaluate(() => (window as Window & { voiceTestStreams?: MediaStream[] }).voiceTestStreams!
.at(-1)!.getAudioTracks()[0]!.getSettings().deviceId)).toBe(selectedDevice.deviceId)
await compareOrRefreshGolden(recordingExpected,
await captureStableAria(page, '[data-composer-card]', scaffold.workspaceCwd), webSnapshotMode())
expect(await page.getByRole('dialog').count()).toBe(0)
@@ -15,8 +15,8 @@
en: d152484eb41ac6b4
zh: 364c80ee00a374ff
/deepseek-ai-dsh-experimental-client-ui-voice-input/use-this-package:
en: 44f25ad14ceda123
zh: 43c4680be42af1cf
en: ff0d98de2970f93d
zh: c234972ea1803afb
/deepseek-ai-dsh-experimental-client-ui-voice-input/understand-the-implementation:
en: b3a7bb1a4f0b3c1e
zh: 233d1c5ea2b16de8
@@ -27,6 +27,8 @@ This optional browser plugin adds an outline microphone icon between the model s
Enable the [voice input bundle](../voice-input-bundle/README.md) from Plugins. After cache inspection, missing models trigger a setup prompt with Go to setup and Later actions. Go to setup opens bundle details; Download and prepare starts installation there. Intact cached models need no prompt or repeated installation. Before recognition is ready, guidance appears only on click, with no microphone tooltip. Clicking the microphone opens the same guidance dialog; its action opens the voice bundle details without recording or downloading. Cache inspection, preparation in progress, failures and disconnected services instead offer navigation to status, progress or error details. The dialog moves focus to its navigation action and keeps Tab traversal inside it; Later or Escape restores the previous focus without changing or submitting the draft. Before preparation, local providers can show their estimated disk, memory and setup-time requirements. The preparation summary starts collapsed; expand it to inspect completed, current and pending steps. Once ready, click the microphone, allow access and click Stop to transcribe. Cancel or Escape discards capture. Recognition feedback stays inside the toolbar. Provider and language preferences are saved from bundle details through the Settings service. Browser microphone access requires HTTPS or loopback and operating-system permission.
Choose **Input device** in bundle details to follow the system default or select a browser microphone. The selected input shows its live level only while the menu is open; selecting another input keeps the menu open for comparison. Closing the menu, hiding the page or disabling the plugin releases the preview, including late permission grants. The browser remembers the device for subsequent recordings independently of Host recognition settings. A missing device or denied permission keeps the menu available for another choice; recording never silently substitutes a different device.
Before downloading or retrying, **Model download source** offers Automatic and the origins advertised by the Host. SenseVoice normally offers Hugging Face and HF-Mirror; a fixed private deployment exposes only its configured origin. Manual selection uses only that source, while Automatic preserves the provider’s comparison and fallback policy. The choice is retained for retries in the current card, without changing deployment configuration or recognition preferences. Source controls are unavailable during preparation and hidden when resources are ready.
-----
@@ -27,6 +27,8 @@ kind: "package-reference"
在插件管理页启用[语音输入 Bundle](../voice-input-bundle/README.zh.md)。缓存检查后,缺少模型时弹出安装引导,可选择“前往安装”或“稍后”。“前往安装”打开 Bundle 详情,再点击“下载并准备”才开始安装。缓存完整时不弹引导,也无需重复安装。识别未就绪时不显示麦克风悬停提示,点击后才打开同一个引导弹窗;点击引导按钮进入语音 Bundle 详情,不会录音或下载。缓存检查、正在准备、准备失败或服务断连时,弹窗改为引导查看状态、进度或错误详情。弹窗将焦点移到导航按钮,并将 Tab 切换约束在弹窗内;选择“稍后”或按 Escape 关闭时恢复此前焦点,不改变或提交草稿。准备前,本地 Provider 可展示磁盘、内存与安装时间的估算。准备摘要默认折叠;展开后可查看已完成、当前和未开始的步骤。就绪后点击麦克风,允许访问,再点击停止以转写。取消或 Escape 丢弃录音。识别反馈留在工具栏内部。Provider 和语言偏好在 Bundle 详情中通过 Settings 服务保存。浏览器麦克风访问需要 HTTPS 或环回地址,以及操作系统权限。
在 Bundle 详情的**输入设备**中选择系统默认或浏览器可用的麦克风。菜单展开时,选中设备旁显示实时输入电平;切换设备后菜单保持展开,便于比较。关闭菜单、隐藏页面或禁用插件会释放预览,包括随后才批准的麦克风授权。浏览器会记住设备选择并用于后续录音,与 Host 的识别设置分别保存。设备缺失或权限被拒绝时,菜单仍可用于重新选择;录音不会静默改用其他设备。
下载或重试前,**模型下载源**提供自动选择及 Host 公布的源。SenseVoice 默认提供 Hugging Face 和 HF-Mirror;固定私有部署仅显示配置的源。手动选择只使用指定源,自动选择沿用提供方的比较与回退策略。当前卡片中的重试会保留选择,不修改部署配置或识别偏好。准备期间不能改源,资源就绪后隐藏源选择器。
-----
@@ -0,0 +1,93 @@
/** Device selection owns microphone previews only while its portaled menu is open. */
import { useEffect, useRef, useState } from 'react'
import { IconCheckOutlineRegular, IconChevronDownOutlineRegular, Menu, StateDot, Tooltip } from '@deepseek-ai/dsh-client-ui-primitives'
import type { InjectFace, PropsLocale } from '@deepseek-ai/dsh-client-ui-slots'
import type { VoiceInputInjected } from './VoiceInput.tsx'
import { RecordingError, type Recording } from './audio.ts'
import type { MicrophoneDevice } from './microphone-device.ts'
import { NS } from './locales.ts'
import css from './VoiceInput.module.css'
function InputLevel({ recording, label }: { recording: Recording; label: string }) {
const meter = useRef<HTMLSpanElement>(null)
useEffect(() => {
const bars = Array.from((meter.current as HTMLSpanElement).children)
let frame: number
const draw = (): void => {
const decibels = 20 * Math.log10(Math.max(recording.amplitude(), 0.00001))
const count = Math.ceil(Math.max(0, Math.min(1, (decibels + 60) / 60)) * bars.length)
for (const [index, bar] of bars.entries()) bar.setAttribute('data-active', String(index < count))
frame = requestAnimationFrame(draw)
}
draw()
return () => { cancelAnimationFrame(frame) }
}, [recording])
return <span ref={meter} className={css.inputLevel} role="img" aria-label={label}>
{Array.from({ length: 8 }, (_, index) => <span key={index} />)}
</span>
}
/** Render the selected input and preview its measured level without recording audio. */
export function MicrophonePicker({ useMicrophoneDevice, selectMicrophone, createRecording, t }:
Pick<InjectFace<VoiceInputInjected>, 'useMicrophoneDevice' | 'selectMicrophone' | 'createRecording'> & PropsLocale<typeof NS>) {
const selected = useMicrophoneDevice(value => value)
const [open, setOpen] = useState(false), [devices, setDevices] = useState<MicrophoneDevice[]>([])
const [preview, setPreview] = useState<Recording>(), [error, setError] = useState(''), [revision, setRevision] = useState(0)
useEffect(() => {
if (!open) return
const capture = createRecording(), media = (navigator as Partial<Navigator>).mediaDevices
const lifetime = new AbortController()
const release = (): void => { void capture.dispose().catch(() => undefined) }
setPreview(undefined); setError('')
const fail = (failure: unknown): void => {
if (lifetime.signal.aborted) return
release(); setPreview(undefined)
setError(failure instanceof RecordingError ? t(failure.kind) : t('failed', {
message: failure instanceof Error ? failure.message : String(failure),
}))
}
const refresh = async (): Promise<void> => {
if (!media) throw new RecordingError('unavailable')
const inputs = (await media.enumerateDevices()).filter(device => device.kind === 'audioinput' && device.deviceId !== '' && device.deviceId !== 'default')
if (!lifetime.signal.aborted) setDevices(inputs.map((device, index) => ({ id: device.deviceId,
label: device.label || t('unnamedMicrophone', { number: String(index + 1) }) })))
}
void (async () => {
try {
await capture.preview(fail)
if (!lifetime.signal.aborted) setPreview(capture)
} catch (failure) { fail(failure) }
if (lifetime.signal.aborted) return
try { await refresh() } catch (failure) { fail(failure) }
})()
const changed = (): void => { setRevision(value => value + 1) }
const hidden = (): void => { if (document.hidden) setOpen(false) }
media?.addEventListener('devicechange', changed, { signal: lifetime.signal })
document.addEventListener('visibilitychange', hidden, { signal: lifetime.signal })
return () => {
lifetime.abort()
release()
}
}, [open, selected.id, revision, createRecording, t])
const choices = [{ id: '', label: t('systemMicrophone') }, ...devices]
const title = selected.id === '' ? t('systemMicrophone') : selected.label
return <div className={css.deviceRow}>
<span>{t('inputDevice')}</span>
<Menu open={open} onClose={() => { setOpen(false) }} portal autoFocus dense selection="fill"
className={css.deviceAnchor} listClassName={css.deviceMenu} selectedId={selected.id}
items={choices.map(device => ({ id: device.id,
icon: <span className={css.deviceCheck}>{device.id === selected.id && <IconCheckOutlineRegular />}</span>,
label: <span className={css.deviceOption}>
<Tooltip label={device.label} portal><span className={css.deviceName}>{device.label}</span></Tooltip>
{device.id === selected.id && (preview ? <InputLevel recording={preview} label={t('inputLevel')} />
: !error && <StateDot state="ongoing" />)}
</span>,
}))}
onSelect={(id) => { selectMicrophone(choices.find(item => item.id === id) as MicrophoneDevice) }}
footer={error ? [{ id: 'error', type: 'label', text: error }] : []}
anchor={<button type="button" className={css.deviceTrigger} aria-label={t('inputDevice')}
aria-haspopup="menu" aria-expanded={open} onClick={() => { setOpen(value => !value) }}>
<span className={css.deviceName}>{title}</span><IconChevronDownOutlineRegular />
</button>} />
</div>
}
@@ -4,6 +4,7 @@ import { Button, DisclosureRow, IconChevronDownOutlineRegular, StateDot, type St
import type { InjectFace, PropsLocale } from '@deepseek-ai/dsh-client-ui-slots'
import type { SpeechPreparationState, SpeechProviderId, SpeechProviderView, SpeechSelectionPatch } from '@deepseek-ai/dsh-experimental-speech-to-text/types'
import type { VoiceInputInjected } from './VoiceInput.tsx'
import { MicrophonePicker } from './MicrophonePicker.tsx'
import { NS } from './locales.ts'
import css from './VoiceInput.module.css'
@@ -129,7 +130,7 @@ export function PreparationCard({ provider, connected, prepare, cancelPreparatio
/** Recognition preferences and preparation cards shared by plugin details and Settings. */
export function VoicePreparation({ useSpeechReadiness, ...props }: Pick<InjectFace<VoiceInputInjected>,
'useSpeechReadiness' | 'configure' | 'prepare' | 'cancelPreparation'> & PropsLocale<typeof NS>) {
'useSpeechReadiness' | 'configure' | 'prepare' | 'cancelPreparation' | 'createRecording' | 'useMicrophoneDevice' | 'selectMicrophone'> & PropsLocale<typeof NS>) {
const readiness = useSpeechReadiness(value => value), catalog = readiness.catalog
const [saving, setSaving] = useState(false), [error, setError] = useState('')
const configure = async (patch: SpeechSelectionPatch): Promise<void> => {
@@ -143,6 +144,7 @@ export function VoicePreparation({ useSpeechReadiness, ...props }: Pick<InjectFa
}
return <div>
{catalog && <div className={css.preferences}>
<MicrophonePicker {...props} />
<label>{props.t('provider')}<select value={catalog.selection.providerId} disabled={!readiness.connected || saving}
onChange={(event) => { void configure({ providerId: event.target.value as SpeechProviderId }) }}>
{catalog.providers.map(provider => <option key={provider.id} value={provider.id}>{provider.name}</option>)}
@@ -37,3 +37,17 @@
.setupEstimate dt { color: var(--dsw-alias-label-tertiary); }
.setupEstimate dd { margin: 0; }
.setupEstimate small { color: var(--dsw-alias-label-tertiary); font-size: 12px; }
.deviceRow { display: grid; grid-template-columns: 100px minmax(120px, 280px); align-items: center; gap: 12px; }
.deviceAnchor { display: block; min-width: 0; }
.deviceTrigger { display: flex; align-items: center; justify-content: space-between; gap: 12px; width: 100%; min-width: 0; border: 0.5px solid var(--dsw-alias-border-l1); border-radius: 6px; padding: 6px 10px; background: transparent; color: inherit; font: inherit; cursor: pointer; }
.deviceMenu { width: 320px; }
.deviceCheck { display: inline-flex; width: 16px; }
.deviceOption { display: flex; align-items: center; justify-content: space-between; gap: 12px; min-width: 0; }
.deviceName { overflow: hidden; text-overflow: ellipsis; white-space: nowrap; min-width: 0; }
.inputLevel { display: inline-flex; align-items: center; gap: 3px; flex: none; }
.inputLevel > span { width: 3px; height: 12px; border-radius: 1px; background: var(--dsw-alias-border-l1); }
.inputLevel > span[data-active='true'] { background: var(--dsw-alias-state-success-primary); }
@media (max-width: 480px) {
.deviceRow, .preferences label { grid-template-columns: minmax(0, 1fr); gap: 6px; }
}
@@ -6,6 +6,7 @@ import type { TranscriptionRequest } from '@deepseek-ai/dsh-experimental-api-spe
import type { SpeechPreparationOptions, SpeechProviderId, SpeechSelection, SpeechSelectionPatch, Transcript } from '@deepseek-ai/dsh-experimental-speech-to-text/types'
import type { RemoteResult } from '@deepseek-ai/dsh-typert-protocol'
import { RecordingError, audioBase64, type Recording } from './audio.ts'
import type { MicrophoneDevice } from './microphone-device.ts'
import type { SpeechReadiness } from './readiness.ts'
import { Waveform } from './Waveform.tsx'
import { VoiceSetupDialog } from './VoiceSetupDialog.tsx'
@@ -19,6 +20,8 @@ export interface VoiceInputActions {
openSettings: () => void
/** @returns one microphone operation owned by the plugin lifecycle. */
createRecording: () => Recording
/** Remember the browser microphone used by previews and subsequent recordings. */
selectMicrophone: (device: MicrophoneDevice) => void
transcribe: (request: TranscriptionRequest, signal: AbortSignal) => Promise<RemoteResult<Transcript>>
prepare: (providerId: SpeechProviderId, options?: SpeechPreparationOptions) => Promise<void>
cancelPreparation: (providerId: SpeechProviderId) => Promise<void>
@@ -27,7 +30,7 @@ export interface VoiceInputActions {
/** Entry-injected Host readiness and microphone operations. */
export interface VoiceInputInjected extends VoiceInputActions {
hooks: { speechReadiness: HostObservable<SpeechReadiness> }
hooks: { speechReadiness: HostObservable<SpeechReadiness>; microphoneDevice: HostObservable<MicrophoneDevice> }
}
/** Composer-owned expansion and draft actions; preferences stay in plugin settings. */
@@ -2,7 +2,7 @@
/** Capture failure whose message is localized by the caller. */
export class RecordingError extends Error {
constructor(readonly kind: 'unavailable' | 'permission' | 'empty' | 'cancelled' | 'interrupted') { super(kind); this.name = 'RecordingError' }
constructor(readonly kind: 'unavailable' | 'permission' | 'empty' | 'cancelled' | 'interrupted' | 'deviceMissing') { super(kind); this.name = 'RecordingError' }
}
/**
@@ -48,21 +48,25 @@ export class Recording {
private readonly lifetime = new AbortController()
private disposal: Promise<void> | undefined
constructor(private readonly onDispose: () => void) {}
constructor(private readonly onDispose: () => void, private readonly deviceId = '') {}
/**
* Acquire the microphone for this recording.
* @param onError - receives failures during capture, before asynchronous resource release finishes.
* @returns after capture starts; a cancelled permission grant immediately releases its tracks.
* Acquire the selected microphone without recording or playing audio.
* @param onError - receives device interruptions before resource release completes.
* @returns after the level analyser starts; cancelled grants release their tracks immediately.
*/
async start(onError?: (error: RecordingError) => void): Promise<void> {
async preview(onError?: (error: RecordingError) => void): Promise<void> {
const devices = (navigator as Partial<Navigator>).mediaDevices
if (!devices || typeof MediaRecorder === 'undefined') throw new RecordingError('unavailable')
if (!devices) throw new RecordingError('unavailable')
let stream: MediaStream
try {
stream = await devices.getUserMedia({ audio: { echoCancellation: true, noiseSuppression: true }, video: false })
stream = await devices.getUserMedia({ audio: { echoCancellation: true, noiseSuppression: true,
...(this.deviceId === '' ? {} : { deviceId: { exact: this.deviceId } }) }, video: false })
} catch (error) {
if (error instanceof DOMException && error.name === 'NotAllowedError') throw new RecordingError('permission')
if (error instanceof DOMException) {
if (error.name === 'NotAllowedError') throw new RecordingError('permission')
if (error.name === 'NotFoundError' || error.name === 'OverconstrainedError') throw new RecordingError('deviceMissing')
}
throw error
}
if (this.lifetime.signal.aborted) { stream.getTracks().forEach((track) => { track.stop() }); throw new RecordingError('cancelled') }
@@ -72,19 +76,37 @@ export class Recording {
this.analyser = this.context.createAnalyser()
this.analyser.fftSize = this.samples.length
this.context.createMediaStreamSource(stream).connect(this.analyser)
this.recorder = new MediaRecorder(stream)
for (const track of stream.getTracks()) track.addEventListener('ended', () => {
this.interrupted(onError)
}, { signal: this.lifetime.signal })
} catch (error) { await this.dispose(); throw error }
}
/**
* Acquire the selected microphone and record audio for transcription.
* @param onError - receives failures during capture, before asynchronous resource release finishes.
* @returns after capture starts; a cancelled permission grant immediately releases its tracks.
*/
async start(onError?: (error: RecordingError) => void): Promise<void> {
if (typeof MediaRecorder === 'undefined') throw new RecordingError('unavailable')
await this.preview(onError)
if (this.lifetime.signal.aborted) throw new RecordingError('cancelled')
try {
this.recorder = new MediaRecorder(this.stream as MediaStream)
this.recorder.ondataavailable = (event) => { if (!this.lifetime.signal.aborted && event.data.size > 0) this.chunks.push(event.data) }
this.recorder.onerror = () => {
if (this.lifetime.signal.aborted) return
void this.dispose().catch(() => undefined)
try { onError?.(new RecordingError('interrupted')) } catch (error) {
console.error('Speech recording error handler failed', error)
}
}
this.recorder.onerror = () => { this.interrupted(onError) }
this.recorder.start()
} catch (error) { await this.dispose(); throw error }
}
private interrupted(onError: ((error: RecordingError) => void) | undefined): void {
if (this.lifetime.signal.aborted) return
void this.dispose().catch(() => undefined)
try { onError?.(new RecordingError('interrupted')) } catch (error) {
console.error('Speech recording error handler failed', error)
}
}
/**
* Read the live microphone signal.
* @returns the measured RMS level, or zero outside capture.
@@ -82,6 +82,11 @@ export const zh = {
discard: '丢弃识别文字',
transcribingShort: '识别中…', wakingShort: '唤醒中…',
provider: '识别服务', language: '识别语言',
inputDevice: '输入设备',
systemMicrophone: '系统默认',
unnamedMicrophone: '麦克风 {number}',
inputLevel: '麦克风输入电平',
deviceMissing: '麦克风不可用,请选择其他输入设备',
auto: '自动识别', zh: '中文', en: '英语', yue: '粤语', ja: '日语', ko: '韩语',
start: '开始录音', stop: '停止并识别', cancel: '取消', insert: '插入文字',
loading: '正在读取识别服务…', requesting: '请允许使用麦克风…', recording: '正在录音…',
@@ -178,6 +183,11 @@ export const en: Record<VoiceKey, string> = {
discard: 'Discard transcript',
transcribingShort: 'Transcribing…', wakingShort: 'Waking…',
provider: 'Recognition service', language: 'Recognition language',
inputDevice: 'Input device',
systemMicrophone: 'System default',
unnamedMicrophone: 'Microphone {number}',
inputLevel: 'Microphone input level',
deviceMissing: 'Microphone unavailable. Choose another input device.',
auto: 'Detect automatically', zh: 'Chinese', en: 'English', yue: 'Cantonese', ja: 'Japanese', ko: 'Korean',
start: 'Start recording', stop: 'Stop and transcribe', cancel: 'Cancel', insert: 'Insert text',
loading: 'Loading recognition services…', requesting: 'Allow microphone access to continue…', recording: 'Recording…',
@@ -0,0 +1,20 @@
/** Browser-local microphone preference shared by settings previews and dictation. */
import { createSnapshotStore, type SnapshotStore } from '@deepseek-ai/dsh-client-store'
/** An empty id follows the system input; labels remain readable before a permission grant. */
export interface MicrophoneDevice {
readonly id: string
readonly label: string
}
/**
* Restore the browser preference independently of Host recognition settings.
* @returns validated device selection, persisted through the shared Client store.
*/
export function createMicrophoneDeviceStore(): SnapshotStore<MicrophoneDevice> {
const store = createSnapshotStore<MicrophoneDevice>({ id: '', label: '' }, { persist: { name: 'dsh.voice-input.microphone' } })
const saved: unknown = store.getSnapshot()
if (!saved || typeof saved !== 'object' || !('id' in saved) || typeof saved.id !== 'string'
|| !('label' in saved) || typeof saved.label !== 'string') store.set({ id: '', label: '' })
return store
}
@@ -8,6 +8,7 @@ import type {} from '@deepseek-ai/dsh-client-ui-conversation/client'
import type { TypertRemoteContribution } from '@deepseek-ai/dsh-typert-protocol'
import { VoiceInput, type VoiceInputInjected } from './VoiceInput.tsx'
import { Recording } from './audio.ts'
import { createMicrophoneDeviceStore } from './microphone-device.ts'
import { en, NS, zh } from './locales.ts'
import type {} from '@deepseek-ai/dsh-client-ui-plugin-manager/client'
import { observeReadiness } from './readiness.ts'
@@ -19,14 +20,16 @@ export const inject = ['remote', 'slots', 'locale', 'pluginNavigation']
function registerUi(ctx: Context): void {
ctx.effect(() => ctx.locale.register(NS, { zh, en }))
const recordings = new Set<Recording>()
const microphone = createMicrophoneDeviceStore()
const readiness = observeReadiness(ctx)
ctx.effect(() => readiness.dispose)
ctx.effect(() => async () => { await Promise.all([...recordings].map(recording => recording.dispose())) })
const actions: VoiceInputInjected = {
openSettings: () => { ctx.pluginNavigation.openBundle('@deepseek-ai/dsh-experimental-voice-input-bundle') },
hooks: { speechReadiness: readiness.state },
hooks: { speechReadiness: readiness.state, microphoneDevice: microphone },
selectMicrophone: (device) => { microphone.set(device) },
createRecording: () => {
const recording = new Recording(() => { recordings.delete(recording) })
const recording = new Recording(() => { recordings.delete(recording) }, microphone.getSnapshot().id)
recordings.add(recording)
return recording
},
@@ -36,13 +36,16 @@ export function captureFixture(options: { empty?: boolean; recorderError?: boole
const offline = vi.fn(function Offline(_channels: number, _frames: number, _rate: number) {
return { destination: {}, createBufferSource: () => ({ buffer: null, connect() {}, start() {} }), startRendering: rendering }
})
vi.stubGlobal('navigator', { mediaDevices: { getUserMedia: async () => ({ getTracks: () => [{ stop: trackStop }] }) } })
const track = new EventTarget()
const stream = { getTracks: () => [Object.assign(track, { stop: trackStop })] }
const getUserMedia = vi.fn(async (_constraints: MediaStreamConstraints) => stream)
vi.stubGlobal('navigator', { mediaDevices: { getUserMedia } })
vi.stubGlobal('MediaRecorder', Recorder)
vi.stubGlobal('AudioContext', function Audio() { return { state: 'running', close, decodeAudioData: decoding,
createMediaStreamSource: () => ({ connect: vi.fn() }),
createAnalyser: () => ({ fftSize: 256, getFloatTimeDomainData: (buffer: Float32Array) => { buffer.fill(0.25) } }),
} })
vi.stubGlobal('OfflineAudioContext', offline)
return { recording: new Recording(disposed), trackStop, close, disposed, decoding, rendering, offline,
return { recording: new Recording(disposed), getUserMedia, stream, track, trackStop, close, disposed, decoding, rendering, offline,
failRecorder: () => { failRecorder() } }
}
@@ -134,3 +134,56 @@ it('contains a failing interruption callback and still releases the microphone',
expect(report).toHaveBeenCalledWith('Speech recording error handler failed', error)
} finally { report.mockRestore(); await b.recording.dispose() }
})
it('previews an exact microphone without creating a recorder and releases it when disconnected', async () => {
const b = captureFixture({ constructError: true }), failure = vi.fn()
const preview = new Recording(() => {}, 'usb-microphone')
try {
await preview.preview(failure)
expect(b.getUserMedia).toHaveBeenCalledWith({ audio: { deviceId: { exact: 'usb-microphone' },
echoCancellation: true, noiseSuppression: true }, video: false })
expect(preview.amplitude()).toBe(0.25)
b.track.dispatchEvent(new Event('ended'))
expect(failure).toHaveBeenCalledWith(expect.objectContaining({ kind: 'interrupted' }))
expect(b.trackStop).toHaveBeenCalledOnce()
expect(preview.amplitude()).toBe(0)
} finally { await preview.dispose() }
})
it.each(['NotFoundError', 'OverconstrainedError'])('reports an unavailable selected microphone on %s without falling back', async (name) => {
const b = captureFixture(), recording = new Recording(() => {}, 'removed')
b.getUserMedia.mockRejectedValueOnce(new DOMException('missing device', name))
try {
await expect(recording.start()).rejects.toMatchObject({ kind: 'deviceMissing' })
expect(b.getUserMedia).toHaveBeenCalledOnce()
} finally { await recording.dispose() }
})
it('releases the microphone if audio analysis cannot initialize', async () => {
const b = captureFixture()
vi.stubGlobal('AudioContext', function AudioUnavailable() { throw new Error('audio unavailable') })
try {
await expect(b.recording.preview()).rejects.toThrow('audio unavailable')
expect(b.trackStop).toHaveBeenCalledOnce()
} finally { await b.recording.dispose() }
})
it('preserves browser device-busy errors', async () => {
const b = captureFixture(), error = new DOMException('busy', 'NotReadableError')
b.getUserMedia.mockRejectedValueOnce(error)
try { await expect(b.recording.preview()).rejects.toBe(error) } finally { await b.recording.dispose() }
})
it('does not create a recorder after cancellation at preview completion', async () => {
const b = captureFixture(), ready = Promise.withResolvers<undefined>(), proceed = Promise.withResolvers<undefined>()
const preview = b.recording.preview.bind(b.recording)
vi.spyOn(b.recording, 'preview').mockImplementation(async (onError) => { await preview(onError); ready.resolve(undefined); await proceed.promise })
const starting = b.recording.start(), rejected = expect(starting).rejects.toMatchObject({ kind: 'cancelled' })
try {
await ready.promise
await b.recording.dispose()
proceed.resolve(undefined)
await rejected
expect(b.trackStop).toHaveBeenCalledOnce()
} finally { proceed.resolve(undefined); await b.recording.dispose() }
})
@@ -0,0 +1,182 @@
// @vitest-environment jsdom
/** Menu-only capture, selected input routing, and late permissions use real audio ownership. */
import { act, cleanup, fireEvent, render, screen, waitFor, within } from '@testing-library/react'
import { bindSnapshotSelector, makeTranslate } from '@deepseek-ai/dsh-client-test-runtime'
import { zh as commonZh } from '@deepseek-ai/dsh-client-locale/src/locales/zh.ts'
import { afterEach, beforeEach, expect, it, vi } from 'vitest'
import { MicrophonePicker } from '../src/client/MicrophonePicker.tsx'
import { createMicrophoneDeviceStore } from '../src/client/microphone-device.ts'
import { Recording } from '../src/client/audio.ts'
import { zh } from '../src/client/locales.ts'
import { captureFixture } from './audio-fixture.client.ts'
const t = makeTranslate(zh, commonZh)
beforeEach(() => {
localStorage.clear()
vi.stubGlobal('requestAnimationFrame', vi.fn(() => 1))
vi.stubGlobal('cancelAnimationFrame', vi.fn())
})
afterEach(() => { cleanup(); localStorage.clear(); vi.unstubAllGlobals() })
function fixture() {
const audio = captureFixture(), selection = createMicrophoneDeviceStore()
const enumerateDevices = vi.fn(async () => [
{ kind: 'audioinput', deviceId: 'default', label: 'Default' },
{ kind: 'audioinput', deviceId: 'usb', label: 'USB microphone' },
{ kind: 'audioinput', deviceId: 'internal', label: 'Built-in microphone' },
{ kind: 'audiooutput', deviceId: 'speakers', label: 'Speakers' },
])
const media = Object.assign(new EventTarget(), { getUserMedia: audio.getUserMedia, enumerateDevices })
vi.stubGlobal('navigator', { mediaDevices: media })
const recordings: Recording[] = []
const props = { t, useMicrophoneDevice: bindSnapshotSelector(selection),
selectMicrophone: (device: { id: string; label: string }) => { selection.set(device) },
createRecording: () => {
const capture = new Recording(() => {}, selection.getSnapshot().id)
recordings.push(capture)
return capture
} }
const view = render(<MicrophonePicker {...props} />)
return { ...audio, ...view, media, enumerateDevices, recordings, props, selection }
}
async function open(): Promise<void> {
fireEvent.click(screen.getByRole('button', { name: zh.inputDevice }))
await screen.findByRole('menuitem', { name: 'USB microphone' })
}
it('captures only while open, moves the meter to the selected input, and remembers it for recording', async () => {
const b = fixture()
try {
expect(b.getUserMedia).not.toHaveBeenCalled()
await open()
expect(screen.queryByRole('menuitem', { name: 'Speakers' })).toBeNull()
expect(within(screen.getByRole('menuitem', { name: new RegExp(zh.systemMicrophone) })).getByRole('img', { name: zh.inputLevel })).toBeTruthy()
fireEvent.click(screen.getByRole('menuitem', { name: 'USB microphone' }))
await waitFor(() => { expect(b.getUserMedia).toHaveBeenLastCalledWith({ audio: {
echoCancellation: true, noiseSuppression: true, deviceId: { exact: 'usb' },
}, video: false }) })
expect(b.trackStop).toHaveBeenCalledOnce()
await waitFor(() => { expect(within(screen.getByRole('menuitem', { name: /USB microphone/ })).getByRole('img', { name: zh.inputLevel })).toBeTruthy() })
expect(screen.getAllByRole('img', { name: zh.inputLevel })).toHaveLength(1)
expect(document.querySelectorAll('[data-active="true"]').length).toBeGreaterThan(0)
fireEvent.keyDown(screen.getByRole('menuitem', { name: /USB microphone/ }), { key: 'Escape' })
await waitFor(() => { expect(screen.queryByRole('menu')).toBeNull() })
expect(b.trackStop).toHaveBeenCalledTimes(2)
expect(screen.queryByRole('img', { name: zh.inputLevel })).toBeNull()
expect(createMicrophoneDeviceStore().getSnapshot()).toEqual({ id: 'usb', label: 'USB microphone' })
const recording = b.props.createRecording()
await recording.start()
expect(b.getUserMedia.mock.calls.at(-1)?.[0].audio).toMatchObject({ deviceId: { exact: 'usb' } })
} finally { b.unmount(); await Promise.all(b.recordings.map(recording => recording.dispose())) }
})
it('releases a permission grant arriving after the menu closes', async () => {
const b = fixture(), granted = Promise.withResolvers<typeof b.stream>()
b.getUserMedia.mockReturnValueOnce(granted.promise)
try {
fireEvent.click(screen.getByRole('button', { name: zh.inputDevice }))
fireEvent.click(screen.getByRole('button', { name: zh.inputDevice }))
await act(async () => { granted.resolve(b.stream) })
expect(b.trackStop).toHaveBeenCalledOnce()
expect(screen.queryByRole('img', { name: zh.inputLevel })).toBeNull()
} finally { b.unmount(); await Promise.all(b.recordings.map(recording => recording.dispose())) }
})
it('keeps device choices available after permission denial and refreshes hot-plugged inputs', async () => {
const b = fixture()
b.getUserMedia.mockRejectedValueOnce(new DOMException('denied', 'NotAllowedError'))
try {
await open()
expect(screen.getByText(zh.permission)).toBeTruthy()
fireEvent.click(screen.getByRole('menuitem', { name: 'USB microphone' }))
await screen.findByRole('img', { name: zh.inputLevel })
b.enumerateDevices.mockResolvedValueOnce([{ kind: 'audioinput', deviceId: 'new', label: 'New microphone' }])
b.getUserMedia.mockRejectedValueOnce(new DOMException('unplugged', 'NotFoundError'))
act(() => { b.media.dispatchEvent(new Event('devicechange')) })
await screen.findByRole('menuitem', { name: 'New microphone' })
expect(screen.getByText(zh.deviceMissing)).toBeTruthy()
expect(b.trackStop).toHaveBeenCalledOnce()
expect(screen.queryByRole('img', { name: zh.inputLevel })).toBeNull()
} finally { b.unmount(); await Promise.all(b.recordings.map(recording => recording.dispose())) }
})
it.each([null, 'usb', { id: 3, label: 'USB' }, { id: 'usb' }])('ignores invalid persisted microphone fields: %j', (saved) => {
localStorage.setItem('dsh.voice-input.microphone', JSON.stringify(saved))
expect(createMicrophoneDeviceStore().getSnapshot()).toEqual({ id: '', label: '' })
})
it('closes on page hiding and releases tracks even when AudioContext closure fails', async () => {
const b = fixture()
const hidden = vi.spyOn(document, 'hidden', 'get')
try {
await open()
hidden.mockReturnValue(false)
fireEvent(document, new Event('visibilitychange'))
expect(screen.getByRole('menu')).toBeTruthy()
b.close.mockRejectedValueOnce(new Error('device closed'))
hidden.mockReturnValue(true)
fireEvent(document, new Event('visibilitychange'))
await waitFor(() => { expect(screen.queryByRole('menu')).toBeNull() })
expect(b.trackStop).toHaveBeenCalledOnce()
} finally {
hidden.mockRestore(); b.unmount()
await Promise.allSettled(b.recordings.map(recording => recording.dispose()))
}
})
it.each([new Error('device busy'), 'device busy'])('shows capture failures without preventing another device choice: %s', async (failure) => {
const b = fixture()
b.getUserMedia.mockRejectedValueOnce(failure)
b.enumerateDevices.mockResolvedValueOnce([{ kind: 'audioinput', deviceId: 'unnamed', label: '' }])
try {
fireEvent.click(screen.getByRole('button', { name: zh.inputDevice }))
await screen.findByRole('menuitem', { name: '麦克风 1' })
expect(screen.getByText('语音识别失败:device busy')).toBeTruthy()
} finally { b.unmount(); await Promise.all(b.recordings.map(recording => recording.dispose())) }
})
it('reports unavailable browser capture and enumeration failures inside the menu', async () => {
const b = fixture()
try {
vi.stubGlobal('navigator', {})
fireEvent.click(screen.getByRole('button', { name: zh.inputDevice }))
await screen.findByText(zh.unavailable)
fireEvent.click(screen.getByRole('button', { name: zh.inputDevice }))
vi.stubGlobal('navigator', { mediaDevices: b.media })
b.enumerateDevices.mockRejectedValueOnce(new Error('enumeration failed'))
fireEvent.click(screen.getByRole('button', { name: zh.inputDevice }))
await screen.findByText('语音识别失败:enumeration failed')
} finally { b.unmount(); await Promise.all(b.recordings.map(recording => recording.dispose())) }
})
it('ignores device enumeration that settles after the menu closes', async () => {
const b = fixture(), result = Promise.withResolvers<Awaited<ReturnType<typeof b.enumerateDevices>>>()
b.enumerateDevices.mockReturnValueOnce(result.promise)
try {
fireEvent.click(screen.getByRole('button', { name: zh.inputDevice }))
await waitFor(() => { expect(b.enumerateDevices).toHaveBeenCalledOnce() })
fireEvent.click(screen.getByRole('button', { name: zh.inputDevice }))
await act(async () => { result.resolve([{ kind: 'audioinput', deviceId: 'late', label: 'Late microphone' }]) })
expect(screen.queryByText('Late microphone')).toBeNull()
expect(b.trackStop).toHaveBeenCalledOnce()
} finally { b.unmount(); await Promise.all(b.recordings.map(recording => recording.dispose())) }
})
it('does not publish a preview when the component unmounts at acquisition completion', async () => {
const b = fixture(), acquired = Promise.withResolvers<undefined>(), release = Promise.withResolvers<undefined>()
const createRecording = b.props.createRecording
b.props.createRecording = () => {
const capture = createRecording(), preview = capture.preview.bind(capture)
vi.spyOn(capture, 'preview').mockImplementation(async (onError) => { await preview(onError); acquired.resolve(undefined); await release.promise })
return capture
}
b.rerender(<MicrophonePicker {...b.props} />)
try {
fireEvent.click(screen.getByRole('button', { name: zh.inputDevice }))
await act(async () => { await acquired.promise })
b.unmount()
await act(async () => { release.resolve(undefined) })
expect(b.trackStop).toHaveBeenCalledOnce()
expect(b.enumerateDevices).not.toHaveBeenCalled()
} finally { release.resolve(undefined); await Promise.all(b.recordings.map(recording => recording.dispose())) }
})
@@ -9,6 +9,7 @@ import { afterEach, expect, it, vi } from 'vitest'
import { PreparationCard, VoicePreparation } from '../src/client/PreparationCard.tsx'
import type { VoiceInputProps } from '../src/client/VoiceInput.tsx'
import type { SpeechReadiness } from '../src/client/readiness.ts'
import { Recording } from '../src/client/audio.ts'
import { zh } from '../src/client/locales.ts'
afterEach(() => { cleanup(); vi.useRealTimers() })
@@ -117,6 +118,8 @@ it('persists settings and reports disconnection in the detail card', async () =>
} })
const configure = vi.fn<VoiceInputProps['configure']>(async () => {})
const props = { useSpeechReadiness: bindSnapshotSelector(store), configure, t,
useMicrophoneDevice: bindSnapshotSelector(createSnapshotStore({ id: '', label: '' })),
selectMicrophone: vi.fn(), createRecording: () => new Recording(() => {}),
prepare: vi.fn(async () => {}), cancelPreparation: vi.fn(async () => {}) }
const view = render(<VoicePreparation {...props} />)
fireEvent.change(screen.getByLabelText(zh.provider), { target: { value: 'cloud' } })
@@ -32,6 +32,7 @@ function fixture(recording?: Recording) {
async () => ({ ok: true, value: transcript }))
const props: VoiceInputProps = { sessionId: 'one' as SessionId, inputActions, transcribe, locked: false, onActiveChange: vi.fn(),
openSettings: vi.fn(), prepare: vi.fn(async () => {}), cancelPreparation: vi.fn(async () => {}), configure: vi.fn(async () => {}),
selectMicrophone: vi.fn(), useMicrophoneDevice: bindSnapshotSelector(createSnapshotStore({ id: '', label: '' })),
useSpeechReadiness: bindSnapshotSelector(readiness), createRecording: () => recording ?? capture,
t: makeTranslate(zh, commonZh) }
const view = render(<VoiceInput {...props} />)
@@ -0,0 +1,6 @@
- menu:
- menuitem "System default"
- menuitem "Fake Audio Input 1 Microphone input level":
- text: Fake Audio Input 1
- img "Microphone input level"
- menuitem "Fake Audio Input 2"