Is there a way to map pitch from an audio to another one ? #3024
Replies: 1 comment 5 replies
-
|
Were you able to find any solution ? :) |
Beta Was this translation helpful? Give feedback.
-
|
Unfortunately not :( |
Beta Was this translation helpful? Give feedback.
-
|
The concept you mentioned is our senior design project and we are having hard time finding tools to accomplish. If there are any other resources, we would be grateful if you can share. Thank you for your response, have a great week :) |
Beta Was this translation helpful? Give feedback.
-
|
@digdemyildiz I do have a smilar research project as well. Please leave a comment, if there's any development. |
Beta Was this translation helpful? Give feedback.
-
|
When I paused this "project" I found a Spotify tool that can extract a MIDI pitch file from an audio (https://github.com/spotify/basic-pitch) and the idea was to get original voice pitch file and then apply it to the generated TTS audio. It must be possible because a little french startup is working on a tool they call Lipitt that can apparently work on songs (some examples https://twitter.com/search?q=lipitt&src=typed_query&f=live ). |
Beta Was this translation helpful? Give feedback.
-
|
Thanks a lot. We will review the resources. Unfortunately, we do not have a prior experience in NLP or Audio Processing, that's why we are searching for already implemented models, but we were not able find one too :( |
Beta Was this translation helpful? Give feedback.
-
The concept is : take a song, translate lyrics, generate voice with these lyrics, apply pitch from original song (isolated voice) to the TTS generated voice, put back music.
It would probably be adjusted manually in some way because for example french is often longer to say something compared to english, but there surely has to be some starting way.
Beta Was this translation helpful? Give feedback.
All reactions