Deploy Whisper ASR
Whisper ASR 1.10: speech-to-text API with faster-whisper, private network.
Just deployed
/data
Deploy and Host Whisper ASR on Railway
Whisper ASR Webservice wraps OpenAI's Whisper speech recognition models in a simple HTTP API. You upload an audio or video file and get back a transcript or translation as text, JSON, SRT or VTT subtitles. It supports the faster-whisper engine, which runs well on ordinary CPUs, and language detection.
About Hosting Whisper ASR
This template runs the CPU image onerahmet/openai-whisper-asr-webservice:v1.10.0 with the faster-whisper engine and the base model in int8. The API has no authentication, so it stays on the private network. Your backend posts files to http://whisper.railway.internal:9000/asr. The model is downloaded from Hugging Face on first boot and cached on a Railway volume, so later deploys start faster. The image is about 1.7 GB, and the first start takes a few minutes. Larger models improve accuracy but need more memory and time per file, so pick one that suits your plan. CPU threads are capped at four.
Common Use Cases
- Transcribing voice notes, calls or meetings uploaded to your app
- Generating subtitles for videos as SRT or VTT
- Private speech-to-text without sending audio to a third party
Dependencies for Whisper ASR Hosting
onerahmet/openai-whisper-asr-webservice:v1.10.0(CPU image)- A Railway volume at
/datafor the model cache
Deployment Dependencies
Implementation Details
| Service | Source | Networking | Storage |
|---|---|---|---|
| whisper | onerahmet/openai-whisper-asr-webservice:v1.10.0 | private only, port 9000 | volume at /data |
| Variable | Default | Purpose |
|---|---|---|
ASR_MODEL | base | tiny, base, small, medium, large-v3, turbo, or .en variants |
ASR_ENGINE | faster_whisper | Also openai_whisper or whisperx |
ASR_QUANTIZATION | int8 | Weight precision on CPU |
OMP_NUM_THREADS | 4 | CPU threads |
WHISPER_URL | http://:9000 | Base URL for your backend |
Transcribe from a service in the project:
curl -F audio_file=@meeting.mp3 "$WHISPER_URL/asr?task=transcribe&output=json"
Use output=srt or output=vtt for subtitles and task=translate to translate into English. The Swagger UI at /docs lists every option.
This is a community-maintained deployment package and does not imply affiliation with or endorsement by OpenAI or the Whisper ASR Webservice project.
Why Deploy Whisper ASR on Railway?
Railway is a singular platform to deploy your infrastructure stack. Railway will host your infrastructure so you don't have to deal with configuration, while allowing you to vertically and horizontally scale it.
By deploying Whisper ASR on Railway, you are one step closer to supporting a complete full-stack application with minimal burden. Host your servers, databases, AI agents, and more on Railway.
Template Content
