• Audio

  • text to speech - thoughts?

Ive been having a bit of fun with pink trombone on salt, and also the speech mode of Mutable Instruments Plaits.

got me thinking, that it could be fun to have a text to speech synth running on salt/pepper.
(basic idea would be to store 'phrases' on usb stick, so easy to add new and edit)

I wondered if anyone had any thoughts on this:
i see two candidates for the underlying library
FLITE - https://github.com/festvox/flite
eSpeakNG - https://github.com/espeak-ng

both open source, and been around a long time, and so i think low cpu.
(flite esp. is was running on WinCE !)
both seem to compile on linux, and i think are ok on bela 🙂

an 'easy' way to make this work is to get bela to precompile the waveforms i(into wav files) n a background process, then just play them back in the patch...

but it probably be more 'fun' to have this done in the patch, so we get a bit more control from cv/analog inputs etc.

anyone had any thoughts on this?

Im currently compiling flite on bela.. so thought i come here whilst im waiting for compile to finish 😉

    thetechnobear

    nice! very interesting. what about an lpc implementation? not direct text to speech, but maybe realtime en/decoding could be accomplished. there are opensource tools to create lpc data from recorded speech.

    i have a set of axoloti objects (adapted from arduino code) that can read lpc data and output speech 🙂

      thetechnobear ot me thinking, that it could be fun to have a text to speech synth running on salt/pepper.

      I expected people would start sequencing the PinkTrombone with all sorts of modulation sources to make it speak, but I am not aware of anyone doing that 🙁

          giuliomoro actually this is what I’ve been playing with - though at the moment it’s better described as ‘throat singing’ 😉

          all started with a discussion regarding the new iOS. app by tim exile - https://endlesss.fm , been sharing a few salt samples - but hope will get this into some jams real soon 🙂

          I’ll post an example once I’ve got it singing nicely.

          btw : it also works nicely as a kind of resonator/filter - I like the way you split off the vocal tract to audio in - great idea !

          .... that’s also what got me thinking about robot voices

              thetechnobear btw : it also works nicely as a kind of resonator/filter - I like the way you split off the vocal tract to audio in - great idea !

              That was exactly the purpose! There is also a CV out to control an external oscillator so that it can be summed into the filter input (the two inputs are normalled together) 🙂

              thetechnobear r described as ‘throat singing’ 😉

              that's a nice start

              thetechnobear all started with a discussion regarding the new iOS. app by tim exile -

              Tim actually got to play the Pink Trombone on the early Salt prototype at Superbooth 2017 ... maybe he remembers that

                    thetechnobear First go at voicing on salt with pink trombone

                    Not sure what it is saying but it sounds good!

                      a month later