-
Implemented the model based on Deep Speech 2: End-to-End Speech Recognition in English and Mandarin
-
The code uses Spectrogram for feature extractor
-
Supports training on your own languages
- Python 3.5 or later
- Keras (https://keras.io/)
- librosa (pip install librosa)
Run the requirements.txt file
pip install -r requirements.txt