Repo: https://github.com/erew123/alltalk_tts (V2 lives on the alltalkbeta branch) · License: AGPL-3.0
GUI: ✅ Web admin panel · Cloning: ✅ XTTS / F5-TTS · Italian: ✅ (XTTS, or a Piper Italian model)
Default port: 7851
A long-running Windows-first TTS server (standalone or as a text-generation-webui extension) that bundles several engines — Coqui XTTS, Piper, F5-TTS, Coqui VITS, Parler-TTS — behind one server with a Web admin panel, finetuning tools and a bulk generator.
Prerequisites: Git, Microsoft C++ Build Tools (needed to compile some Python packages), Windows SDK, and eSpeak NG.
Clone the V2 branch:
git clone -b alltalkbeta https://github.com/erew123/alltalk_tts
cd alltalk_tts
atsetup.bat
http://localhost:7851).alloy, echo, fable, nova, onyx,
shimmer) maps to a real voice/clone you choose.| App field | Value |
|---|---|
| Endpoint | http://<pc-ip>:7851 |
| API key | empty, or the key set in AllTalk’s OpenAI settings |
| model | tts-1 (ignored by AllTalk — it uses the engine selected in its panel) |
| voice | alloy … shimmer — each mapped server-side to your voice/clone |
| lang | it |
model is required but ignored: engine switching happens in the AllTalk
UI, not via the API model field. One OpenAI voice = one mapped clone; if you
want several Android voices, map several voice names in the panel.