END TO END ASR TRANSFORMER

• 本项目基于transformer 6*encoder+6*decoder的基本结构构造的端到端的语音识别系统

Model

Instructions

• 1.数据准备:
  • 自行下载数据，遵循文件结构如下：

├── data
│   ├── train
│   ├── dev
│   ├── test

• 2.数据预处理：
  • 运行prepare_data.py对数据进行预处理, 获得整个词表，每个样本音频的mel-scale-spectrogram，文本的token-ids
• 3.模型训练：
  • 运行train_transformer.py --ngpus 8进行transformer网络的训练. 该网络输入mel-scale-spectrogram, 输出token-ids
• 4.模型推理：
  • 运行evlauate.py在dev/test上测试准确率

Acknowledgements

• GitHub ESPET
• GitHub Speech-Transformer-tf2.0
• Github speech-transformer

Reference

• Ashish Vaswani et al. “Attention Is All You Need” (2017).
• Abdel-rahman Mohamed et al. “Transformers with convolutional context for ASR” arXiv: Computation and Language (2019).
• Albert Zeyer et al. “Improved Training of End-to-end Attention Models for Speech Recognition” Conference of the International Speech Communication Association (2018).

GitHub

View Github