Nickrus
DF Admirer
I think a major problem is pasting a generated soundtrack with a celebrity voice. It actually should happen in several steps:
Less useful but might be interesting:
- Collect original celebrity soundtracks WITH alignes subtitles.
- Split soundtracks into separate words and expressions.
- Create a tesaurus (dictionary, words library).
- Generate sentences based on the dictionary (SOFTWARE MISSING)
- Paste the soundtrack into the target video.
- eSpeak - reading aloud texts, might be possible to use own voice.
- ModelTalker - some software able to being trained for a specific voice.
- Voicery - commercial appliance, own voice creation possible.
- Mivoc - commercal service, own voice creation possible.
- Cerevoc - commercal service, own voice creation possible.
Less useful but might be interesting:
- Adobe Voco - they made a proof-of-concept version but never released it into open market.
- Yamaha Vocaloid is a mature product but it's specialized to generate singing tracks, and the voice libraries are pre-buit, pre-shipped. No official way to prepare own voice library.
- maryTTS - abandoned TTS open source project