Note

This project was previously named faster-whisper-server. I've decided to change the name from faster-whisper-server, as the project has evolved to support more than just transcription.

Speaches

speaches is an OpenAI API-compatible server supporting streaming transcription, translation, and speech generation. Speach-to-Text is powered by faster-whisper and for Text-to-Speech piper and Kokoro are used. This project aims to be Ollama, but for TTS/STT models.

Try it out on the HuggingFace Space

See the documentation for installation instructions and usage: https://speaches-ai.github.io/speaches/

Features:

• GPU and CPU support.

• Deployable via Docker Compose / Docker

• Highly configurable

• OpenAI API compatible. All tools and SDKs that work with OpenAI's API should work with speaches.

• Streaming support (transcription is sent via SSE as the audio is transcribed. You don't need to wait for the audio to fully be transcribed before receiving it).

  • LocalAgreement2 (paper | original implementation) algorithm is used for live transcription.

• Live transcription support (audio is sent via websocket as it's generated).

• Dynamic model loading / offloading. Just specify which model you want to use in the request and it will be loaded automatically. It will then be unloaded after a period of inactivity.

• Text-to-Speech via kokoro(Ranked #1 in the TTS Arena) and piper models.

• Coming soon: Audio generation (chat completions endpoint) | OpenAI Documentation

  • Generate a spoken audio summary of a body of text (text in, audio out)
  • Perform sentiment analysis on a recording (audio in, text out)
  • Async speech to speech interactions with a model (audio in, audio out)

• Coming soon: Realtime API | OpenAI Documentation

Please create an issue if you find a bug, have a question, or a feature suggestion.

Demo

Streaming Transcription

TODO

Speech Generation

2025-01-12_13-20-58.webm

Live Transcription (using WebSockets)

demo.mp4