luma/src/modules/tts-modes/azure.ts

import { createHash, randomBytes } from 'crypto';
import { TTSModule, TTSResponse } from '../tts';

import * as https from 'https';

import { WebSocket } from 'ws'
import { Logger } from '../../utils/log';

const CLIENT_TOKEN = "6A5AA1D4EAFF4E9FB37E23D68491D6F4";
const AZURE_ENDPOINT = "speech.platform.bing.com";

const READALOUD_PATH = `/consumer/speech/synthesize/readaloud`
const WEBSOCKET_URL = `wss://${AZURE_ENDPOINT}${READALOUD_PATH}/edge/v1?TrustedClientToken=${CLIENT_TOKEN}`;
const VOICES_PATH = `${READALOUD_PATH}/voices/list?TrustedClientToken=${CLIENT_TOKEN}`;

const CHROME_VERSION = '138.0.7204.157';
const SEC_VERSION = `1-${CHROME_VERSION}`;

const USER_AGENT = `Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/${CHROME_VERSION.split('.')[0]}.0.0.0 Safari/537.36 Edg/${CHROME_VERSION.split('.')[0]}.0.0.0`;
const WIN_EPOCH = 11644473600;

const WS_RECONNECT_DELAY = 2000;
const MAX_RECONNECT_ATTEMPTS = 5;

interface PendingRequest {
	resolve: (value: TTSResponse) => void;
	reject: (reason: Error) => void;
	audioBuff: Buffer[];
}

class AzureTTS implements TTSModule {
	private voices: Array<string> | undefined = undefined;

	public name: string = 'Azure';
	public defaultVoice: string = 'en-US-AvaNeural';

	private ready: boolean = false;
	private readyPromise: Promise<void> | null = null;
	private readyResolve: (() => void) | null = null;

	private ws: WebSocket | undefined = undefined;
	private reconnectAttempts: number = 0;
	private reconnectTimer: NodeJS.Timeout | null = null;
	private isReconnecting: boolean = false;

	private log: Logger;
	// Map keyed by X-RequestId
	private pendingRequests: Map<string, PendingRequest> = new Map();
	constructor() {
		this.log = new Logger('Azure TTS');
		this.initializeConnection();
	}

	async getVoices(): Promise<Array<string> | undefined> {
		if (this.voices)
			return this.voices;

		const options: https.RequestOptions = {
			hostname: AZURE_ENDPOINT,
			path: `${VOICES_PATH}&Sec-MS-GEC=${this.genSecToken()}&Sec-MS-GEC-Version=${SEC_VERSION}`,
			method: 'GET',
			headers: {
				'Pragma': 'no-cache',
				'Cache-Control': 'no-cache',
				'User-Agent': USER_AGENT,
				"Accept-Encoding": "gzip, deflate, br",
				"Accept-Language": "en-US,en;q=0.9",
				"Authority": "speech.platform.bing.com",
				"Sec-CH-UA": `" Not;A Brand";v="99", "Microsoft Edge";v="${CHROME_VERSION.split('.')[0]}", "Chromium";v="${CHROME_VERSION.split('.')[0]}"`,
				"Sec-CH-UA-Mobile": "?0",
				"Accept": "*/*",
				"Sec-Fetch-Site": "none",
				"Sec-Fetch-Mode": "cors",
				"Sec-Fetch-Dest": "empty",
			}
		};

		return new Promise((resolve) => {
			const req = https.request(options, (res) => {
				const chunks: Buffer[] = [];
				res.on('data', (chunk) => chunks.push(chunk));
				res.on('end', () => {
					const body = Buffer.concat(chunks).toString();
					this.voices = JSON.parse(body).map((v: any) => v.ShortName)
					resolve(this.voices);
				});
				req.on('error', (err) => {
					throw err;
				});
				res.on('aborted', () => {
					throw new Error('Response aborted')
				});
			});
			req.end();
		});
	}

	async generate(voice: string, text: string): Promise<TTSResponse> {
		await this.readyPromise;
		if (!this.ready || !this.ws) return { error: 'Not initialized' };

		const reqId = randomBytes(16).toString('hex');
		const lang = voice.split('-').slice(0, 2).join('-');

		return new Promise((resolve, reject) => {
			this.pendingRequests.set(reqId, { resolve, reject, audioBuff: [] });

			const headers = `X-RequestId:${reqId}\r\nContent-Type:application/ssml+xml\r\nPath:ssml\r\n\r\n`;
			const ssml = `<speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis" xml:lang="${lang}"><voice name="${voice}"><prosody rate="default" pitch="default">${this.escapeXml(text)}</prosody></voice></speak>`;

			this.ws?.send(headers + ssml, (err) => {
				if (err) {
					this.pendingRequests.delete(reqId);
					reject(err);
				}
			});
		});
	}

	canBeUsed(): boolean {
		return true;
	}

	private initializeConnection(): void {
		this.ready = false;
		this.readyPromise = new Promise((resolve) => {
			this.readyResolve = resolve;
			this.connect();
		});
	}

	private connect(): void {
		const url = `${WEBSOCKET_URL}&Sec-MS-GEC=${this.genSecToken()}&Sec-MS-GEC-Version=${SEC_VERSION}`;

		this.ws = new WebSocket(url, {
			host: 'speech.platform.bing.com',
			origin: 'chrome-extension://jdiccldimpdaibmpdkjnbmckianbfold',
			headers: {
				'Pragma': 'no-cache',
				'User-Agent': USER_AGENT,
			}
		});

		this.ws.on('open', () => {
			this.log.verbose('WebSocket open');
			this.reconnectAttempts = 0;
			this.isReconnecting = false;

			const config = `Content-Type:application/json; charset=utf-8\r\nPath:speech.config\r\n\r\n
            {
                "context": {
                    "synthesis": {
                        "audio": {
                            "metadataoptions": { "sentenceBoundaryEnabled": "false", "wordBoundaryEnabled": "true" },
                            "outputFormat": "audio-24khz-48kbitrate-mono-mp3"
                        }
                    }
                }
            }`;
			this.ws?.send(config.trim());
			this.ready = true;
			this.readyResolve?.();
		});

		this.ws.on('message', (data: Buffer, isBinary: boolean) => {
			this.handleIncomingMessage(data, isBinary);
		});

		this.ws.on('close', (code/*, reason*/) => {
			this.ready = false;
			this.log.verbose(`WS Closed: ${code}`);
			this.rejectAllPending(new Error("Connection closed"));
			this.scheduleReconnect();
		});

		this.ws.on('error', (err) => {
			this.log.error('WS Error:', err);
		});
	}

	private scheduleReconnect() {
		if (this.reconnectAttempts >= MAX_RECONNECT_ATTEMPTS) return;

		const delay = WS_RECONNECT_DELAY * Math.pow(2, this.reconnectAttempts++);
		setTimeout(() => this.connect(), delay);
	}

	private handleIncomingMessage(data: Buffer, isBinary: boolean) {
		const message = data.toString();
		const reqId = message.match(/X-RequestId:(.*?)\r\n/)?.[1];
		if (!reqId) return;

		const request = this.pendingRequests.get(reqId);
		if (!request) return;

		if (isBinary) {
			const separator = 'Path:audio\r\n';
			const index = data.indexOf(separator);
			if (index !== -1) {
				request.audioBuff.push(data.subarray(index + separator.length));
			}
		} else {
			if (message.includes('Path:turn.end')) {
				request.resolve({ data: Buffer.concat(request.audioBuff) });
				this.pendingRequests.delete(reqId);
			} else if (message.includes('Path:turn.error') || message.includes('Path:error')) {
				request.reject(new Error("Azure synthesis error"));
				this.pendingRequests.delete(reqId);
			}
		}
	}

	private rejectAllPending(err: Error) {
		for (const [id, req] of this.pendingRequests) {
			req.reject(err);
			this.pendingRequests.delete(id);
		}
	}

	private genSecToken(): string {
		const ticks = BigInt(Math.floor((Date.now() / 1000) + Number(WIN_EPOCH))) * 10000000n
		const roundedTicks = ticks - (ticks % 3000000000n)

		const strToHash = `${roundedTicks}${CLIENT_TOKEN}`

		const hash = createHash('sha256')
		hash.update(strToHash, 'ascii')

		return hash.digest('hex').toUpperCase()
	}

	private escapeXml(unsafe: string): string {
		return unsafe.replace(/[<>&"']/g, (c) => {
			switch (c) {
				case '<': return '&lt;'
				case '>': return '&gt;'
				case '&': return '&amp;'
				case '"': return '&quot;'
				case "'": return '&apos;'
				default: return c
			}
		})
	}
}

export default new AzureTTS();