Jun 13Vibe with Hermes Agent — Bengaluru · RSVP
ToolsMCPBlogResearchCommunityStar on GitHub
All Tools
W
OtherFreeOpen Source

WENET

Production-ready end-to-end ASR

Apache-2.0

ABOUT

Deploying speech recognition in production — especially with streaming requirements for live transcription and voice interfaces — often means juggling separate non-streaming accuracy and streaming latency models. WeNet unifies streaming and non-streaming ASR in a single two-pass architecture trained end-to-end with PyTorch, offering pretrained checkpoints and a lightweight runtime so teams can go from data to deployed transcription without complex model ensembles.

INSTALL
pip install git+https://github.com/wenet-e2e/wenet.git

INTEGRATION GUIDE

1. Deploy real-time streaming speech recognition for voice assistants and live captioning 2. Train end-to-end ASR models on custom domains with the PyTorch training pipeline 3. Run on-device or server-side inference with WeNet's lightweight runtime 4. Benchmark streaming versus non-streaming accuracy for production trade-off decisions

TAGS

speech-recognitionasrstreamingpytorchaudiodeep-learningproduction