# QuentinFuxa/WhisperLiveKit

Real-time, local speech-to-text with streaming ASR, speaker diarization, translation, and OpenAI/Deepgram-compatible APIs.

Repository: https://github.com/QuentinFuxa/WhisperLiveKit
Canonical: https://ross.abutalabs.com/products/whisperlivekit
Language: Python
License: Apache-2.0
License Family: permissive
Topics: python, real-time, speaker-diarization, speech-recognition, speech-to-text, streaming, translation, websocket, whisper, automatic-speech-recognition, fastapi, pytorch
Last push: 2026-08-24T04:38:41+00:00

## Health v2 (maintenance only)
Score: 87/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 95, longevity 44
- inputs: {"age_days": 622, "days_push": 9, "days_rel": 32, "gap_med": 10, "n_releases_24m": 32}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 10962, forks 1133 (observed 2026-08-28T04:10:44.853713+00:00)

## What it is
WhisperLiveKit is a self-hosted, ultra-low-latency real-time speech-to-text pipeline built on state-of-the-art simultaneous speech research (SimulWhisper, WhisperStreaming, Streaming Sortformer). It provides streaming transcription, speaker diarization, and simultaneous translation via WebSocket and OpenAI/Deepgram-compatible APIs.

## Use cases
- transcribe live audio in real time locally
- add real-time captions to a meeting or stream
- diarize speakers in a live conversation
- translate speech simultaneously across 200 languages
- self-host an OpenAI-compatible transcription API
- build a voice assistant with low-latency ASR

## When to choose
- you need low-latency streaming transcription rather than batch file transcription
- you want everything running locally without sending audio to cloud APIs
- you need speaker diarization or simultaneous translation alongside transcription
- you want an OpenAI/Deepgram-compatible endpoint you can drop into existing clients

## When to avoid
- you only need to transcribe pre-recorded audio files offline
- you have no GPU and need high-throughput transcription
- you want a fully managed hosted service with no infrastructure to run

## Facets
- artifact type: library
- maturity: active
- function: speech-recognition, websocket, http-server, api-framework, nlp
- domain: speech-processing, machine-learning, self-hosted
- platform: python, cross-platform
- tags: whisper, streaming-asr, speaker-diarization, simultaneous-translation, fastapi, pytorch, transcription, openai-compatible-api, natural-language-processing, real-time, gpu, web-server, docker

## Member repositories
- QuentinFuxa/WhisperLiveKit (main) score 87

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:44.853713+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:17:17.397910+00:00, confidence not recorded.
  - readme: https://github.com/QuentinFuxa/WhisperLiveKit (fetched 2026-08-28T04:10:44.853713+00:00, sha 14ce6dd5910d)
  - registry_pypi: https://pypi.org/pypi/whisperlivekit/json (fetched 2026-08-29T08:15:53.958533+00:00, sha 515ed8e9d6a3)
- Data as of 2026-08-30T08:39:29.467469+00:00.