adi-gov-tw/Taiwan-Tongues-ASR-CE
Taiwan Tongues ASR CE 是一個開源語音辨識（Automatic Speech Recognition, ASR）模型專案，專為台灣多元語言環境設計。 本模型支援 國語、台語、客語與英語，提供本地多語混合語音辨識，讓開發者與資訊服務業者可運用此開源模型進行 ASR 模型訓練、微調與發展在地化應用，以低成本、高效率進行 ASR 語音應用落地與智慧服務創新。 本專案為數位發展部數位產業署「114年數位產業跨域軟體基盤系統建置案」之實證成果之一，旨在推動台灣語音技術開源生態，協助資訊服務業者強化智慧應用能量，落實在地 AI 技術自主發展，由台灣大哥大執行與維護。
Built on Whisper architecture with faster-whisper/CTranslate2 inference backend for optimized CPU/GPU performance, the project provides both batch file transcription (HTTP task queue with TSV/SRT/DIA subtitle output) and real-time WebSocket streaming with voice activity detection. Training pipeline integrates HuggingFace Transformers with modular dataset handling, supporting multi-language fine-tuning via `train_asr.py` and evaluation through character error rate (CER) metrics, with pre-configured models deployable as FastAPI services.
Stars
29
Forks
12
Language
Python
License
—
Category
Last pushed
Dec 30, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/voice-ai/adi-gov-tw/Taiwan-Tongues-ASR-CE"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
YuanGongND/whisper-at
Code and Pretrained Models for Interspeech 2023 Paper "Whisper-AT: Noise-Robust Automatic Speech...
huggingface/distil-whisper
Distilled variant of Whisper for speech recognition. 6x faster, 50% smaller, within 1% word error rate.
KevKibe/African-Whisper
🚀 Framework for seamless fine-tuning of Whisper model on a multi-lingual dataset and deployment to prod.
sandy1990418/ChineseTaiwaneseWhisper
This repository focuses on leveraging OpenAI's Whisper model for speech recognition in Chinese...
ga642381/Taiwanese-Whisper
fine-tune Whipser model for Taiwanese speech recognition