typed text reconstruction based on audio only
- data collection
- data preprocessing
- supervised learning
- deployment
- sync audio chunks with keypress
- create mel spectrogram
- convert keypress to model targets
- store results
name | function |
---|---|
analytics.py | Analyze collected data distribution |
collect.py | Data collection pipeline |
models.py | Pytorch deep learning models |
preprocess.py | Data preprocessing, creating single data file |
test.py | Testing validation results |
train.py | Used to train models from models.py |
util.py | Utility functions |