A technology that I am surprised that doesn't exist yet is altering one person's voice to match another's. Both people read the same calibration text, or even different texts which both cover most of the phonemes, and then it seems a simple matter to map over the phonemes.
Dub the audio track of a movie from one language to another, preserving the original actor's voice. The intermediate steps (not done automatically) include translation, then a stand-in voice actor speaking the translation, then the audio of the translation being adjusted from the voice of voice actor to the original actor.
Of course, such technology makes it difficult to trust that an audio recording is authentic.
Famous calibration texts:
The North Wind and the Sun
These may be problematic if the target language includes sounds that do not occur in English, though interpolation or extrapolation may be sufficient.
No comments :
Post a Comment