CorentinJ / Real-Time-Voice-Cloning

Clone a voice in 5 seconds to generate arbitrary speech in real-time

Other

52.59k stars 8.79k forks source link

Quality of generated audio #1060

Open aryanpanpalia opened 2 years ago

aryanpanpalia commented 2 years ago

When using a recording from the LibriSpeech downloaded dataset, a good ratio of the generated audio pieces sound good and accurate. However, whenever I record some audio and use that, no matter who the speaker is, all the generated audio pieces sound the same. Is there any way I can fix this, or am I not understanding how to use this tool correctly? I've seen others on YouTube using the tool the same way I am and the resulting audio clips sound far better than my own,

raccoonML commented 2 years ago

Provide an example? It may be the quality of your recordings or your microphone setup, like distance from speaker or recording environment.

scatterp2 commented 2 years ago

437 has a dropbox link which dont exist so kinda hard to reproduce