Prosody Distribution
This git repository contains three modules.
1. bigram.py:
A naive probabilistic model that prints out the conditional probability of tone symbol. In order to execute, following procedure.
1) Extract Boston Radio Speech Corpus
2) Place "bu\_radio" directory in current directory.
3) sh reset.sh
4) sh retrieve\_files.sh
5) python3 preprocessing.py
6) python3 bigram.py
2. LSTM.ipynb:
- Prosody detection and generation using pytorch LSTM.
- Nevertheless, this
.ipynb
file contains previous output.
3. BERT.ipynb:
- Prosody detection and generation using BERT-NER.
- Colab Link
Contact me at heehcs@gmail.com if you have any question or comment.