mirror of
https://github.com/CherryHQ/cherry-studio.git
synced 2025-12-25 11:20:07 +08:00
284 lines
9.2 KiB
TypeScript
284 lines
9.2 KiB
TypeScript
import i18n from '@renderer/i18n'
|
||
|
||
import { TTSServiceInterface } from './TTSServiceInterface'
|
||
|
||
// 全局变量来跟踪当前正在播放的语音
|
||
let currentUtterance: SpeechSynthesisUtterance | null = null
|
||
|
||
// 全局变量来跟踪是否正在播放
|
||
export let isEdgeTTSPlaying = false
|
||
|
||
/**
|
||
* Edge TTS服务实现类
|
||
*/
|
||
export class EdgeTTSService implements TTSServiceInterface {
|
||
private edgeVoice: string
|
||
|
||
/**
|
||
* 构造函数
|
||
* @param edgeVoice Edge语音
|
||
*/
|
||
constructor(edgeVoice: string) {
|
||
this.edgeVoice = edgeVoice
|
||
console.log('初始化EdgeTTSService，语音:', edgeVoice)
|
||
}
|
||
|
||
/**
|
||
* 验证参数
|
||
* @throws 如果参数无效，抛出错误
|
||
*/
|
||
private validateParams(): void {
|
||
if (!this.edgeVoice) {
|
||
throw new Error(i18n.t('settings.tts.error.no_edge_voice'))
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 直接播放语音
|
||
* @param text 要播放的文本
|
||
* @returns 是否成功播放
|
||
*/
|
||
private playDirectly(text: string): boolean {
|
||
try {
|
||
// 验证参数
|
||
this.validateParams()
|
||
|
||
// 使用Web Speech API
|
||
if (!('speechSynthesis' in window)) {
|
||
throw new Error(i18n.t('settings.tts.error.browser_not_support'))
|
||
}
|
||
|
||
// 停止当前正在播放的语音
|
||
window.speechSynthesis.cancel()
|
||
if (currentUtterance) {
|
||
currentUtterance = null
|
||
}
|
||
isEdgeTTSPlaying = false
|
||
|
||
// 创建语音合成器实例
|
||
const utterance = new SpeechSynthesisUtterance(text)
|
||
currentUtterance = utterance
|
||
isEdgeTTSPlaying = true
|
||
|
||
// 获取可用的语音合成声音
|
||
const voices = window.speechSynthesis.getVoices()
|
||
console.log('可用的语音合成声音:', voices)
|
||
|
||
// 查找指定的语音
|
||
let selectedVoice = voices.find((v) => v.name === this.edgeVoice)
|
||
|
||
// 如果没有找到指定的语音，尝试使用中文语音
|
||
if (!selectedVoice) {
|
||
console.warn('未找到指定的语音:', this.edgeVoice)
|
||
// 尝试找中文语音
|
||
selectedVoice = voices.find((v) => v.lang === 'zh-CN')
|
||
|
||
if (selectedVoice) {
|
||
console.log('使用替代中文语音:', selectedVoice.name)
|
||
} else {
|
||
// 如果没有中文语音，使用第一个可用的语音
|
||
if (voices.length > 0) {
|
||
selectedVoice = voices[0]
|
||
console.log('使用第一个可用的语音:', selectedVoice.name)
|
||
} else {
|
||
console.warn('没有可用的语音')
|
||
return false
|
||
}
|
||
}
|
||
} else {
|
||
console.log('已选择语音:', selectedVoice.name)
|
||
}
|
||
|
||
// 设置语音
|
||
if (selectedVoice) {
|
||
utterance.voice = selectedVoice
|
||
}
|
||
|
||
// 设置事件处理程序
|
||
utterance.onend = () => {
|
||
console.log('语音合成已结束')
|
||
currentUtterance = null
|
||
isEdgeTTSPlaying = false
|
||
|
||
// 分发一个自定义事件，通知语音合成已结束
|
||
// 这样TTSService可以监听这个事件并重置播放状态
|
||
const event = new CustomEvent('edgeTTSComplete', { detail: { text } })
|
||
document.dispatchEvent(event)
|
||
}
|
||
|
||
utterance.onerror = (errorEvent) => {
|
||
console.error('语音合成错误:', errorEvent)
|
||
currentUtterance = null
|
||
isEdgeTTSPlaying = false
|
||
|
||
// 在错误时也触发结束事件，确保状态更新
|
||
const completeEvent = new CustomEvent('edgeTTSComplete', { detail: { text, error: true } })
|
||
document.dispatchEvent(completeEvent)
|
||
}
|
||
|
||
// 开始语音合成
|
||
window.speechSynthesis.speak(utterance)
|
||
return true
|
||
} catch (error) {
|
||
console.error('直接播放语音失败:', error)
|
||
return false
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 合成语音
|
||
* @param text 要合成的文本
|
||
* @returns 返回音频Blob对象的Promise
|
||
*/
|
||
async synthesize(text: string): Promise<Blob> {
|
||
// 验证参数
|
||
this.validateParams()
|
||
|
||
// 先尝试直接播放
|
||
const playResult = this.playDirectly(text)
|
||
if (playResult) {
|
||
// 如果直接播放成功，返回一个有效的音频Blob
|
||
// 创建一个简单的音频文件，包含一个短暂停
|
||
// 这个文件可以被浏览器正常播放，但实际上不会发出声音
|
||
// 因为我们已经使用Web Speech API直接播放了语音
|
||
const silentAudioBase64 = 'UklGRiQAAABXQVZFZm10IBAAAAABAAEARKwAAIhYAQACABAAZGF0YQAAAAA='
|
||
const silentAudioBuffer = Uint8Array.from(atob(silentAudioBase64), (c) => c.charCodeAt(0))
|
||
return new Blob([silentAudioBuffer], { type: 'audio/wav' })
|
||
}
|
||
|
||
// 如果直接播放失败，尝试录制方法
|
||
console.log('直接播放失败，尝试录制方法')
|
||
try {
|
||
console.log('使用浏览器TTS生成语音，音色:', this.edgeVoice)
|
||
|
||
// 使用Web Speech API
|
||
if (!('speechSynthesis' in window)) {
|
||
throw new Error(i18n.t('settings.tts.error.browser_not_support'))
|
||
}
|
||
|
||
// 停止当前正在播放的语音
|
||
window.speechSynthesis.cancel()
|
||
isEdgeTTSPlaying = false
|
||
|
||
// 创建语音合成器实例
|
||
const utterance = new SpeechSynthesisUtterance(text)
|
||
|
||
// 获取可用的语音合成声音
|
||
const voices = window.speechSynthesis.getVoices()
|
||
console.log('初始可用的语音合成声音:', voices)
|
||
|
||
// 如果没有可用的声音，等待声音加载
|
||
if (voices.length === 0) {
|
||
try {
|
||
await new Promise<void>((resolve) => {
|
||
const voicesChangedHandler = () => {
|
||
window.speechSynthesis.onvoiceschanged = null
|
||
resolve()
|
||
}
|
||
window.speechSynthesis.onvoiceschanged = voicesChangedHandler
|
||
|
||
// 设置超时，防止无限等待
|
||
setTimeout(() => {
|
||
window.speechSynthesis.onvoiceschanged = null
|
||
resolve()
|
||
}, 5000)
|
||
})
|
||
} catch (error) {
|
||
console.error('等待语音加载超时:', error)
|
||
}
|
||
}
|
||
|
||
// 重新获取可用的语音合成声音
|
||
const updatedVoices = window.speechSynthesis.getVoices()
|
||
console.log('更新后可用的语音合成声音:', updatedVoices)
|
||
|
||
// 查找指定的语音
|
||
let selectedVoice = updatedVoices.find((v) => v.name === this.edgeVoice)
|
||
|
||
// 如果没有找到指定的语音，尝试使用中文语音
|
||
if (!selectedVoice) {
|
||
console.warn('未找到指定的语音:', this.edgeVoice)
|
||
// 尝试找中文语音
|
||
selectedVoice = updatedVoices.find((v) => v.lang === 'zh-CN')
|
||
|
||
if (selectedVoice) {
|
||
console.log('使用替代中文语音:', selectedVoice.name)
|
||
} else {
|
||
// 如果没有中文语音，使用第一个可用的语音
|
||
if (updatedVoices.length > 0) {
|
||
selectedVoice = updatedVoices[0]
|
||
console.log('使用第一个可用的语音:', selectedVoice.name)
|
||
} else {
|
||
console.warn('没有可用的语音')
|
||
}
|
||
}
|
||
} else {
|
||
console.log('已选择语音:', selectedVoice.name)
|
||
}
|
||
|
||
// 设置语音
|
||
if (selectedVoice) {
|
||
utterance.voice = selectedVoice
|
||
}
|
||
|
||
// 创建一个Promise来等待语音合成完成
|
||
return await new Promise<Blob>((resolve, reject) => {
|
||
try {
|
||
// 使用AudioContext捕获语音合成的音频
|
||
const audioContext = new (window.AudioContext || (window as any).webkitAudioContext)()
|
||
const audioDestination = audioContext.createMediaStreamDestination()
|
||
const mediaRecorder = new MediaRecorder(audioDestination.stream)
|
||
const audioChunks: BlobPart[] = []
|
||
|
||
mediaRecorder.ondataavailable = (event) => {
|
||
if (event.data.size > 0) {
|
||
audioChunks.push(event.data)
|
||
}
|
||
}
|
||
|
||
mediaRecorder.onstop = () => {
|
||
const audioBlob = new Blob(audioChunks, { type: 'audio/wav' })
|
||
resolve(audioBlob)
|
||
}
|
||
|
||
// 开始录制
|
||
mediaRecorder.start()
|
||
|
||
// 设置语音合成事件
|
||
utterance.onend = () => {
|
||
// 语音合成结束后停止录制
|
||
setTimeout(() => {
|
||
mediaRecorder.stop()
|
||
}, 500) // 等待一下，确保所有音频都被捕获
|
||
}
|
||
|
||
utterance.onerror = (event) => {
|
||
console.error('语音合成错误:', event)
|
||
mediaRecorder.stop()
|
||
reject(new Error('语音合成错误'))
|
||
}
|
||
|
||
// 开始语音合成
|
||
window.speechSynthesis.speak(utterance)
|
||
|
||
// 设置超时，防止无限等待
|
||
setTimeout(() => {
|
||
if (mediaRecorder.state === 'recording') {
|
||
console.warn('语音合成超时，强制停止')
|
||
mediaRecorder.stop()
|
||
}
|
||
}, 10000) // 10秒超时
|
||
} catch (error: any) {
|
||
console.error('浏览器TTS语音合成失败:', error)
|
||
reject(new Error(`浏览器TTS语音合成失败: ${error?.message || '未知错误'}`))
|
||
}
|
||
})
|
||
} catch (error: any) {
|
||
console.error('浏览器TTS语音合成失败:', error)
|
||
// 即使失败也返回一个空的Blob，而不是抛出异常
|
||
// 这样可以避免在UI上显示错误消息
|
||
return new Blob([], { type: 'audio/wav' })
|
||
}
|
||
}
|
||
}
|