search

gukush / audio-watermark-242

Repository for research project about watermarkng audio
3 stars 0 forks source link

Add support for YourTTS voice conversion #10

Open gukush opened 2 weeks ago

gukush commented 2 weeks ago

https://github.com/Edresson/YourTTS

The aim of this task is to analyze solution and integrate it into our voice cloning inventory

MDG369 commented 1 day ago

YourTTS model is included in coqui-ai-tts models, so the implementation works the same as in issue #7.
Example TTS command: tts --model_name tts_models/multilingual/multi-dataset/your_tts --speaker_wav /audio/voice-jewish-1.wav --reference_wav /audio/voice-polish-8.wav --language_idx "en" The converted audio file is of low quality and the speakers are not recognizable.

MDG369 commented 1 day ago

Test in the form of a python script was added in commit cb49316eef41f08fc4999154a762a66bda3b4d0c

gukush commented 7 hours ago

For some reason coqui-ai-tts doesn't utilize this in their API: python yourtts_test.py Traceback (most recent call last): File "/project/tests/yourtts_test.py", line 7, in tts.voice_conversion_to_file(source_wav="/audio/voice-hispanic-1.wav", target_wav="/audio/voice-polish-8.wav", file_path="output.wav") File "/project/coqui-ai-TTS/TTS/api.py", line 377, in voice_conversion_to_file wav = self.voice_conversion(source_wav=source_wav, target_wav=target_wav) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/project/coqui-ai-TTS/TTS/api.py", line 358, in voice_conversion wav = self.voice_converter.voice_conversion(source_wav=source_wav, target_wav=target_wav) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Changed the test so that it calls the CLI app from subprocess in commit 48b84606713bae17d06dd526c72a977e8cc85de7