erew123
7e6e18c6d9
A BETA test/Proof of concept script to compare Speech that was generated in TTS Generator to the original Text and flag issues. Anyone reading this, you are welcome to try it, the readme instructions are in the folder, however, I am not providing support on this while it is BETA/Proof of concept.
34 lines
1.4 KiB
Plaintext
34 lines
1.4 KiB
Plaintext
USAGE NOTES
|
|
Separate Execution: Run this script independently in its own command prompt or terminal window, separate from AllTalk's operational environment.
|
|
|
|
JSON File Requirement: After TTS generation, export your list as ttsList.json and ensure it is located in the same directory as ttsgencomp.py.
|
|
|
|
Iterative Process: Should you need to regenerate TTS for any flagged IDs, repeat the export and script execution process with your updated JSON list.
|
|
|
|
CUDA Dependency: Similar to AllTalk's finetuning, this script requires access to cublas64_11. Ensure your system meets the necessary requirements.
|
|
https://github.com/erew123/alltalk_tts/tree/main?#-important-requirements-cuda-118
|
|
|
|
INSTRUCTIONS:
|
|
1) Ensure your active the Python environment that AllTalk runs in & Ensure you have the CUDA dependency set up (link above)
|
|
|
|
2) Install the required packages using pip:
|
|
|
|
pip install faster-whisper fuzzywuzzy spacy
|
|
|
|
3) Download the language model en_core_web_md for Spacy:
|
|
|
|
python -m spacy download en_core_web_md
|
|
|
|
4) Generate your TTS in the TTS Generator and export the ttsList.JSON to .../alltalk_tts/templates/ttsgencomp/
|
|
|
|
5) Ensure you're in the directory where ttsgencomp.py is located (Replace /path/to/ with the actual path to the .../alltalk_tts/templates/ttsgencomp/ directory):
|
|
|
|
cd /path/to/alltalk_tts/templates/ttsgencomp/
|
|
|
|
6) Run the script using Python:
|
|
|
|
python ttsgencomp.py
|
|
|
|
7) Follow any on screen instructions.
|
|
|