README.md

metadata

title: README
emoji: 📈
colorFrom: pink
colorTo: red
sdk: streamlit
pinned: false
sdk_version: 1.43.2
thumbnail: >-
  https://cdn-uploads.huggingface.co/production/uploads/629e1b71bb6419817ed7566c/jeUU2sPSuMRP9IIqVnufk.png

• GenSEC: Text-based Generative Audio & Speech Recognition with Cascaded ASR-LLMs

  • Task 1: ASR N-best hypotheses correction
  • Task 2: Speaker Tagging from N-best hypotheses
  • Task 3: Emotion Recognition from N-best hypotheses

• Open Source Model

  • Llama-7b pre-training for ASR correction
    • https://huggingface.co/GenSEC-LLM/SLT-Task1-Llama2-7b-HyPo-baseline

• IEEE SLT 2024, References Paper. See below resources for baseline models and datasets.

@inproceedings{yang2024large,
  title={Large language model based generative error correction: A challenge and baselines for speech recognition, speaker tagging, and emotion recognition},
  author={Yang, Chao-Han Huck and Park, Taejin and Gong, Yuan and Li, Yuanchao and Chen, Zhehuai and Lin, Yen-Ting and Chen, Chen and Hu, Yuchen and Dhawan, Kunal and {\.Z}elasko, Piotr and others},
  booktitle={2024 IEEE Spoken Language Technology Workshop (SLT)},
  pages={371--378},
  year={2024},
  organization={IEEE}
}