Open Issues Need Help
View All on GitHubAI Summary: The task is to debug a text preprocessing issue in the Coqui TTS library's Korean Fairseq model. The model's vocabulary file (vocab.txt) contains only Latin characters, causing Korean characters to be discarded during text processing. The solution requires investigating the model's data loading and preprocessing steps to identify why the Korean vocabulary is missing and correcting it to include the necessary Korean ideograms.
πΈπ¬ - a deep learning toolkit for Text-to-Speech, battle-tested in research and production
AI Summary: The task is to debug a bug in the Coqui TTS server's web interface. The bug causes the server to fail when a reference audio file is provided for voice conversion because the JavaScript code incorrectly sets the speaker ID to an empty string. The solution involves modifying either the JavaScript code in `index.html` to correctly handle the `speaker_wav` parameter or adjusting the Python server code in `server.py` to handle empty speaker IDs gracefully.
πΈπ¬ - a deep learning toolkit for Text-to-Speech, battle-tested in research and production