edge-tts.js

function mkssml(text, voice, rate, volume) {
    return (
        "<speak version='1.0' xmlns='http://www.w3.org/2001/10/synthesis' xml:lang='en-US'>" +
        `<voice name='${voice}'><prosody pitch='+0Hz' rate='${rate}' volume='${volume}'>` +
        `${text}</prosody></voice></speak>`
    )
}
function ssmlHeadersPlusData(request_id, timestamp, ssml) {
    return (
        `X-RequestId:${request_id}\r\n` +
        'Content-Type:application/ssml+xml\r\n' +
        `X-Timestamp:${timestamp}Z\r\n` + // This is not a mistake, Microsoft Edge bug.
        `Path:ssml\r\n\r\n` +
        `${ssml}`
    )
}
const text = 'who are you'
const TRUSTED_CLIENT_TOKEN = '6A5AA1D4EAFF4E9FB37E23D68491D6F4'
const WSS_URL =
    'wss://speech.platform.bing.com/consumer/speech/synthesize/' +
    'readaloud/edge/v1?TrustedClientToken=' +
    TRUSTED_CLIENT_TOKEN
const connectId = crypto.randomUUID().replaceAll('-', '')
const ws = new WebSocket(`${WSS_URL}&ConnectionId=${connectId}`)
const date = new Date()
ws.onopen = () => {
    ws.send(
        `X-Timestamp:${date}\r\n` +
            'Content-Type:application/json; charset=utf-8\r\n' +
            'Path:speech.config\r\n\r\n' +
            '{"context":{"synthesis":{"audio":{"metadataoptions":{' +
            '"sentenceBoundaryEnabled":false,"wordBoundaryEnabled":true},' +
            '"outputFormat":"audio-24khz-48kbitrate-mono-mp3"' +
            '}}}}\r\n'
    )
    ws.send(
        ssmlHeadersPlusData(
            connectId,
            date,
            mkssml(text, 'Microsoft Server Speech Text to Speech Voice (en-US, AriaNeural)', '+0%', '+0%')
        )
    )
}
const audioContext = new AudioContext()
let when = audioContext.currentTime + 1
ws.addEventListener('message', async (event) => {
    console.log(event.data)
    if (event.data instanceof Blob) {
        const str = await event.data.text()
        const data = await event.data.slice(str.indexOf('Path:audio\r\n') + 12).arrayBuffer()
        const buffer = await audioContext.decodeAudioData(data)
        const source = audioContext.createBufferSource()
        source.buffer = buffer
        source.connect(audioContext.destination)
        source.start(when)
        when += buffer.duration
    }
})