Jeremy Shi:
- Preprocessin the audio and transciption data (segmentation, uploading, etc.)
- Google Cloud VM GPU set up
- Swig Decoder (adopted from Baidu's DeepSpeech2 on PaddlePaddle)
- Presentation Notebook (live demo)
- Testing Word Error Rate on DASS
- Documentation
- Paper Writing
- Code Review
Ailing Wang
- Building stacked-LSTM model in Keras
- Google Cloud training and testing
- Presentation Notebook (images, theories, etc.)
- Testing Word Error Rate on LibriSpeech
- Documentation
- Paper Writing
- Code Review