Migushthe2nd/MsEdgeTTS

What it is

MsEdgeTTS is a Node‑JS library that lets you generate speech audio by calling Microsoft Edge’s built‑in Read Aloud service (the same service Azure Speech uses). It builds the required SSML, opens a WebSocket to the Edge API, and streams back the synthesized audio.

Why it matters

• No Azure key needed – it re‑uses the free Edge Read Aloud endpoint, so you can get neural‑quality TTS without a cloud subscription.
• Streaming output – audio is returned as a Node stream, which is handy for real‑time playback or piping into other tools.
• Fine‑grained control – you can pick voice, output format, rate, pitch, volume, and even request word‑ and sentence‑boundary metadata.
• Works server‑side – the API now requires an Edge‑style user‑agent, so it runs in Node (or any server‑side runtime) but not in other browsers.

How to use it (quick start)

import {MsEdgeTTS, OUTPUT_FORMAT} from "msedge-tts";

const tts = new MsEdgeTTS();
await tts.setMetadata(
  "en-US-AriaNeural",               // voice
  OUTPUT_FORMAT.WEBM_24KHZ_16BIT_MONO_OPUS // audio codec
);

// Stream audio directly
const {audioStream} = tts.toStream("Hello world!");
audioStream.on('data', chunk => console.log('audio chunk', chunk));

You can also write straight to a file with toFile(), adjust rate/pitch/volume via the optional options object, or supply a custom HTTP/SOCKS proxy agent.

Key features

Feature	Details
Voice selection	Pass any Edge voice identifier (e.g., en-IE-ConnorNeural).
Output formats	Several WebM/Opus, MP3, OGG, etc., defined in OUTPUT_FORMAT.
SSML support	Generates basic SSML with <speak>, <voice>, and <prosody> tags.
Boundary metadata	Optional word‑ and sentence‑boundary events returned as a separate stream or file.
Timeout handling	Configurable connection and synthesis timeouts; emits errors on the audio stream.
Proxy support	Accepts a custom http.Agent (e.g., https-proxy-agent, socks-proxy-agent).
Promise‑based API	All methods return promises; streams emit standard Node events.

Who might use it

• Developers building voice assistants, bots, or IVR systems that need quick TTS without Azure billing.
• Content creators who want to batch‑generate narration files server‑side.
• Researchers prototyping speech‑enabled applications where low‑latency streaming is useful.
• Anyone needing word‑level timing data for lip‑sync or subtitle generation.

---

All information is taken directly from the repository’s README; no additional features are assumed.