nl8590687/ASRT_SpeechRecognition

Health v2 · maintenance only

Adoption not part of the score

8383 stars · 1893 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

ASRT is a deep-learning-based Chinese speech recognition (speech-to-text) system built with TensorFlow/Keras, using CNN, LSTM, attention mechanisms, and CTC loss. It supports training custom acoustic models on standard Chinese speech datasets and running inference on audio files.

Use cases

• transcribe Chinese audio to text
• train a Chinese speech recognition model
• convert Mandarin speech recordings into pinyin and Chinese text
• build a voice-to-text service for Chinese
• fine-tune an ASR model on custom Chinese speech data
• evaluate speech recognition accuracy on Chinese datasets

When to choose

• you need offline, self-hosted Chinese speech recognition
• you want to train or customize an ASR model for Mandarin
• you have GPU hardware and want an open-source research-grade Chinese STT pipeline

When to avoid

• you need multilingual speech recognition beyond Chinese
• you want a production-ready managed ASR API without training infrastructure
• you lack a GPU with 11GB+ memory for training

Facets

library · maturity active

speech-recognition deep-learning machine-learning audio-processing speech-processing machine-learning python windows chinese-speech-recognition speech-to-text ctc cnn keras tensorflow asr natural-language-processing linux gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="nl8590687/ASRT_SpeechRecognition")