How to Deploy ESPnet with Docker: ASR and TTS in Minutes
Step-by-step tutorial to run ESPnet in Docker: pull the image, transcribe audio with a pretrained ASR model, and synthesize speech with TTS.
ESPnet's official Docker image gets you a full speech-processing environment in minutes. Here is how to deploy it, run an ASR model, and generate speech with TTS — using only the espnet/espnet image.
🚀 Want to deploy ESPnet yourself?
Docker configs, system requirements, and installation guides — all on one page.
View ESPnet Tool Page →Step 1: Pull the image and start a container
ESPnet publishes ready-made images on Docker Hub with all dependencies preinstalled. Start an interactive container with GPU passthrough if available:
docker run --gpus all -it --rm -v $(pwd)/data:/data espnet/espnet:latest bash
Step 2: Run ASR on an audio file
Inside the container, ESPnet2's espnet2 command downloads a pretrained model from Hugging Face automatically and transcribes audio:
espnet2-asr-inference --model "espnet/owsm_v1.2" --audio /data/meeting.wav
Step 3: Synthesize speech with TTS
Text-to-speech works the same way. Pick a voice from the model zoo and generate a WAV file in one command:
espnet2-tts-inference --model "espnet/vits_ljspeech" --text "Hello from Run This Ai" --output /data/out.wav
Resource requirements
| Component | Minimum | Recommended |
|---|---|---|
| CPU | 2 cores | 4+ cores |
| RAM | 4 GB | 8 GB |
| GPU | Optional | NVIDIA (CUDA) for fast inference |
🚀 Run ESPnet in production?
Get the ready-to-use docker-compose file and full specs on the tool page.
View ESPnet Tool Page →