Initial standalone dectalk TTS server extracted from the ttstoy bot cog

Self-contained HTTP service with engine, data, start script, systemd unit, and documentation. Runs independently of the Discord bot on its fixed port.
2026-09-14 14:31:15 -05:00
commit b29d04ac2c
10 changed files with 1591 additions and 0 deletions
@@ -0,0 +1,102 @@
+# DECTalk TTS Server
+
+A standalone HTTP server that turns text into speech using authentic DECTalk
+(the Moonbase Alpha voice). It exposes both a simple GET endpoint and a
+MiniMax-compatible POST endpoint, so it can be used directly or as a drop-in
+TTS backend.
+
+This service was extracted from the ttstoy Discord bot cog so it can run on its
+own as a system service, independent of the bot.
+
+## Requirements
+
+- Node.js (v16 or newer recommended)
+- ffmpeg (used to encode output to mp3)
+- npm dependencies: `express`, `dectalk` (installed via `npm install`)
+
+Install ffmpeg on Debian/Ubuntu:
+
+```bash
+sudo apt install ffmpeg
+```
+
+## Running
+
+```bash
+./start.sh
+```
+
+`start.sh` runs `npm install` on first launch and then starts the server. The
+listening port defaults to `33001` and can be overridden with the `PORT`
+environment variable.
+
+You can also run it directly:
+
+```bash
+npm install
+PORT=33001 node server.js
+```
+
+## API
+
+### GET /say
+
+Generate speech and return an mp3.
+
+```
+GET /say?text=Hello%20world
+```
+
+Response: `audio/mpeg` (mp3 bytes).
+
+Voice is controlled inline using DECTalk voice commands (see below).
+
+### POST /v1/t2a_v2
+
+MiniMax-compatible endpoint. Accepts JSON `{ "text": "..." }` and returns a
+JSON body with the audio encoded as hex under `data.audio`.
+
+```bash
+curl -X POST http://127.0.0.1:33001/v1/t2a_v2 \
+  -H 'Content-Type: application/json' \
+  -d '{"text":"Hello from DECTalk"}'
+```
+
+### GET /voices
+
+Returns the list of available DECTalk voices and their inline command codes.
+
+### GET /engine
+
+Returns engine type and version metadata.
+
+### GET /health
+
+Returns `{ "status": "ok" }`.
+
+## Voice commands
+
+Voices are selected inline by prefixing the text with a command code:
+
+| Voice  | Command | Description             |
+|--------|---------|-------------------------|
+| Paul   | `[:np]` | Perfect Paul (default)  |
+| Betty  | `[:nb]` | Beautiful Betty         |
+| Harry  | `[:nh]` | Huge Harry (deep male)  |
+| Frank  | `[:nf]` | Frail Frank (elderly)   |
+| Dennis | `[:nd]` | Doctor Dennis           |
+| Kit    | `[:nk]` | Kit the Kid (child)     |
+| Ursula | `[:nu]` | Uppity Ursula           |
+| Rita   | `[:nr]` | Rough Rita              |
+| Wendy  | `[:nw]` | Whispering Wendy (soft) |
+
+Example:
+
+```
+GET /say?text=[:nh]This is Huge Harry speaking.
+```
+
+## Running as a system service
+
+A systemd unit file is provided (`dectalk-tts-server.service`). See INSTALL.md
+for setup steps.