Speaches
A self-hosted, OpenAI API-compatible server for local speech-to-text, translation and text-to-speech models.
These buttons open the developer's own site, repository or store listing in a new tab. wares.gg does not host downloads.
About Speaches
Speaches is a server that exposes speech models through an OpenAI-compatible API, so tools and SDKs written for OpenAI's audio endpoints can point at your own machine instead. Transcription and translation use faster-whisper, and speech generation uses the Kokoro and Piper models.
Transcription can stream results over SSE while audio is still being processed. Models load on request and unload after a period of inactivity. It runs on CPU or GPU, supports a Realtime API, and is deployed with Docker or Docker Compose.
Strengths
- Drop-in replacement for OpenAI's speech and transcription endpoints
- Streaming transcription over SSE
- Loads and unloads models automatically on demand
- Kokoro and Piper voices for text-to-speech
Limitations
- A server with an API, not an end-user app
- Setup is based on Docker
Details
- Pricing
- FreeFree and open source under the MIT licence.
- License
- MIT
- Developer
- The Speaches contributors
- Platforms
- Self-hosted
- How it runs
- Self-hosted
- Account
- Not required
- Best suited for
- Developers who want local speech-to-text and text-to-speech behind an OpenAI-style API
- Categories
- Local AI tools, AI voice, AI developer tools
- Last verified
- Added
Alternatives to Speaches
Compare allSoftware that can replace Speaches for an important use case, and what changes if you switch.
Kokoro-FastAPI
A Dockerized, OpenAI-compatible server for running the Kokoro-82M text-to-speech model on your own hardware.
Kokoro-FastAPI provides an OpenAI-compatible speech endpoint in prebuilt Docker images, but it covers only Kokoro text-to-speech, not transcription or translation.
AllTalk TTS
A local text-to-speech server with voice cloning, model finetuning, a settings page and a JSON API.
AllTalk TTS is an AGPL-3.0 local text-to-speech server with a JSON API, voice cloning and finetuning, but it does not provide speech-to-text.
Cartesia
An AI voice platform with an API for real-time text-to-speech, speech-to-text and voice agents.
Cartesia is a paid cloud API covering text-to-speech, transcription and voice agents with low latency, replacing local models with a hosted service.
ElevenLabs
An AI voice platform for text to speech, voice cloning, dubbing, speech to text and music.
ElevenLabs offers text to speech, speech to text and voice cloning through an API and SDKs, but it runs only as an online service.
Similar software
Related functionality, not necessarily a direct replacement.
Whisper
An open-source speech recognition model from OpenAI for transcribing and translating audio on your own machine.
whisper.cpp
C/C++ port of OpenAI's Whisper for fast offline speech-to-text on your own hardware.
Piper
A fast neural text-to-speech engine that runs locally, even on modest hardware like a Raspberry Pi.
Kokoro
Small, fast open-weight text-to-speech model you can run locally.
Scriberr
A self-hosted, offline audio transcription app that runs AI speech recognition on your own hardware.
Whishper
A self-hosted web app for transcribing, translating and editing subtitles locally with Whisper models.
