Anthony C. BartmanMar 8
This was quite amusing. BeSTspeech would work well as a wolf simulator.
Show threadSpacedogMar 10
@rommix0 how did you do that?
Show threadAnthony C. BartmanMar 10
@spacepup Simply used the embedded commands to raise the pitch and lower the speech rate. Nothing else.
Show threadSpacedogMar 10
@rommix0 ah right, i forgot there are embedded commands
Show threadAnthony C. BartmanMar 10
@spacepup indeed
Show threadSpacedogMar 10
@rommix0 how you been by the way. Playing a game called rhythm doctor, and might make a level here out of one eof my songs
Show threadAnthony C. BartmanMar 10
@spacepup It's been okay. I had my fair share of ups and downs. I just recently got back on here after dealing with some stuff.
Show threadSpacedogMar 10
@rommix0 well i certainly missed you buddy.
Show threadAnthony C. BartmanMar 10
@spacepup I missed you guys too. Last year was kinda weird honestly. I'm starting to looking forward to things now.
Show threadSpacedogMar 10
@rommix0 nice! I made an hmm based speech synthesizer, and i tried to make another one cfombining the concatination of unit selection but with the inflection modelling of diphone. Would you mind telling me about your tts project you're working on?
Show threadAnthony C. BartmanMar 10

@spacepup Which one? The formant synthesizer one, or the BestSpeech decompilation project since I'm actually working on both right now.

As far as the formant synthesizer goes, it started out as a freelance project meant to be coded in Rust. In recent months, I've been thinking about getting it over to C as well.

My client has been kinda on and off lately. If things don't pan out with him, I'm going full on in on the C version since per contract I can't just put out the Rust code. That way people can easily create SAPI adapters for it when it's all complete.

Show threadSpacedogMar 10
@rommix0 the formant synthesizer
Show threadAnthony C. BartmanMar 10
@spacepup I figure that's what you meant. Nothing to say much of at the moment, but I'll share you the sound demo of where it's at right now.
Show threadBorrisMar 10
@rommix0 @spacepup That sounds like it really wants to be L&H TrueSpeech when it grows up, or at least it thinks it does right now. Maybe it will totally change its mind.
Show threadAnthony C. Bartman
@BorrisInABox @spacepup That's fair. It is still a little rough around the edges. The main issue is the locus movements. I got the articulation theory down, but the method I'm using at the moment makes it still sound a little stiff.
Mar 10 at 4:29amWeb
Show threadBorrisMar 10
@rommix0 @spacepup Yeah. The base formants are pretty similar anyway.
Show threadSpacedogMar 10
@rommix0 @BorrisInABox and the intonation model is also tricky it seams. I mean, i think you are making the intonation model from scratch?
Show threadAnthony C. BartmanMar 10

@spacepup @BorrisInABox Indeed. I came up with an efficient method of generating Fujisaki like pitch contours.

In fact, I shared the python prototype of the pitch model with Tamas a while back.