Whisper ASR Box

Use it when

• Multiple apps or services need speech to text and you want one self-hosted REST endpoint instead of bundling Whisper into each of them
• You need transcripts with word-level timestamps or speaker diarization from audio that cannot leave your infrastructure

What it solves

Not the fit when

• Live low-latency streaming transcription; it processes uploaded files through a REST service
• Environments without Docker or a compatible Python service runtime
• Speaker diarization requires selecting the WhisperX engine
• The quickstart publishes port 9000 and the README documents no authentication layer; keep it loopback/private or add authenticated TLS termination and upload limits before network exposure
• Only process recordings you are authorized to transcribe, especially when enabling speaker diarization
• real-time streaming captioning
• text summarization of transcripts
• hosted transcription without running your own container

Install

docker run -d -p 9000:9000 -e ASR_MODEL=base -e ASR_ENGINE=openai_whisper onerahmet/openai-whisper-asr-webservice:latest (use :latest-gpu with --gpus all for GPU)

Invoke

Open http://localhost:9000 for the Swagger UI and POST audio files to the REST endpoints; select engine and model via ASR_ENGINE and ASR_MODEL environment variables

Alternatives

No reviewed alternatives recorded yet.