-
2024.8: Release the code for “Low-Resourced Speech Recognition for Iu Mien Language via Weakly-Supervised Phoneme-based Multilingual Pre-training” Readme | Paper".
-
2024.6: Release the code for Streaming multi-channel end-to-end (ME2E) ASR. Readme | Paper
-
2024.6: Release the code for "Whistle: Data-Efficient Multilingual and Crosslingual Speech Recognition via Weakly Phonetic Supervision". Readme | Paper
-
2023.5: Release the code for Exploring Energy-based Language Models with Different Architectures and Training Methods for Speech Recognition. Readme | Paper
-
2022.11: Release of v3, including:
- RNN-Transducer training and decoding implementation (Huahuan Zheng).
- Language model (NN and n-gram) training and inference support (Huahuan Zheng).
- LM fusion support for ASR, including Low Order Density Ratio (LODR) for language model integration (Huahuan Zheng).
- CUSIDE implementation for training unified streaming / non-streaming models (Keyu An, Huahuan Zheng and Ziwei Li). Paper | Readme | 中文说明
- Guide to train models on more than 1500 hours of speech data: English | 中文说明
-
2022.05: Release the code for Join Acoustics and Phonology (JoinAP) for Multi/Cross-lingual ASR. Readme | Tutorial | ASRU2021 Paper | Slides | Video
-
2021.07: add support of Deformable TDNN by Keyu An. INTERSPEECH2021 Paper
-
2021.07: add support of Wordpieces by Wenjie Peng. Readme | Paper
-
2021.05: add support of Conformer and SpecAug by Huahuan Zheng. Readme | Paper