Convert Tibetan text into phonemes
Using this tool, you can clean Tibetan text, remove Chinese and English symbols and special symbols, convert Tibetan text into phoneme labels, and generate words and letters labels with more parameter options.
A sample corpus folder is given in this project. Please change the corresponding corpus folder path for formal use.
Usage: the following scripts are run using Python3
python tibetan_label.py --lexicon ./lexicon.txt --corpus ./corpus/
# The phoneme transcription of the corresponding text is generated in the subfolder of corpusParameter description
| parameter | illustration | default | required |
|---|---|---|---|
| --lexicon | Lexicon file path | ./lexicon.txt | True |
| --corpus | Corpus folder path | ./corpus/ | True |
| --words | Generate words annotation | False | False |
| --letters | Generate letters annotation | False | False |
If you want to generate phones, words and letters labels at the same time, please use the following command
python tibetan_label.py --lexicon ./lexicon.txt --corpus ./corpus/ --words True --letters True
The transcript.py script can clean up a single Tibetan text
python transcript.py --source_file ./corpus/file1/file1.txt --target_file ./corpus/file1/file1_clean.txt
The labels.py script can phonemize a single Tibetan text
python3 labels.py --lexicon ./lexicon.txt --phones ./corpus/file1/file1_phones.txt --transcript ./corpus/file1/file1_clean.txt
This project is licensed under the GNU Affero license.
Please indicate the source, do not use for commercial purposes.