With deep learning approaches, yes. They seem to do a great job of turning any sort of features into intelligible speech.
- David On 05/01/19 12:50, Brian Bartholomew via Freetel-codec2 wrote: > Here's a dumb question. Can you analyze the input speech all the way down to > something like phonemes, then just send those? Input is any person talking, > output sounds like the Stephen Hawking speech synth. Bandwidth is tiny, but > output far better than Morse -> text-to-speech because the speaker's brain > already broke it down into phonemes. > > Brian > > > _______________________________________________ > Freetel-codec2 mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/freetel-codec2 > _______________________________________________ Freetel-codec2 mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/freetel-codec2
