humanoid sounds, uncanny valley 

Revisiting some voice physical model demos I did a few years ago. Mostly it's just porting code from Sporth to Runt.

Here's a cyber monk:

patchwerk nodes

36 -0.1 0.1 1 4 0.2 randi randi add mtof
0.1 1 sine 0 1 biscale
0.9
0.9
0.3 1 sine 0 1 biscale
voc dcblock

36 mtof 70 5 eqfil

bdup bdup 0.97 10000 revsc bdrop -14 ampdb mul dcblock add

"chant.wav" wavout bdrop

sr 30 * _compute rep

Follow

humanoid sounds, uncanny valley 

Here's one babbling. It sounds a bit irritated to me...

# ported from the original sporth code september 2020
patchwerk nodes

100 8 metro 0.3 0 maygate 200 mul 0.1 port add

29 40 1 10 0.5 randi randi add

0 1 3 randi
0 1 3 20 1 randi randi
0.7
0
voc

1 metro 0.7 0 maygate 0.03 port mul

bdup jcrev dcblock add

"rant.wav" wavout bdrop

sr 30 * _compute rep

· · brutaldon · 1 · 1 · 4

humanoid sounds, uncanny valley 

@paul woah! I love both of these for different reasons.

On a totally unrelated note I have no idea why I can play audio without problems from firefox, except from mastodon. (I don't mind having a collection of pbatch sketch downloads on my HD tho.)

humanoid sounds, uncanny valley 

@hecanjog
I love playing around with voice synthesis techniques. Great fun. It's one of the few occasions where computer music has a sense of humor, and totally accessible as a concept to many audiences. Trying to jump back into it again.

Browsers and sound are weird. I've never understood the excitement for web audio, for instance. But then again, I use JACK, Linux, and external sound cards, so it never works. For years, I've been told you can somehow make it work with Firefox somehow, but meh.

web audio 

@paul @hecanjog
I have used Emscripten's SDL port so I can write audio callbacks in C. But then you have to jump through a hoop in that you have to start sound from a user action on the page or it doesn't work in Chromium etc.

Then Firefox on Linux uses PulseAudio so you need the PulseAudio/JACK bridge, which can be a bit fiddly.

Looking into miniaudio.h for my current project, it supposedly supports JACK on Linux as well as Web Audio with Emscripten, and other OS/API too.

web audio 

@mathr @paul @hecanjog using miniaudio right now, works a charm on Win10(WASAPI) and Linux(ALSA).

humanoid sounds, uncanny valley 

@paul @hecanjog

I did one piece with a speech synthesis (flite) / speech recognition (pocket sphinx) feedback loop thing - could be fun to do something with less mainstream/straightforward speech synth stuff in it.

maybe even with neural networks or genetic programming or something so it can "learn" how to say things.

humanoid sounds, uncanny valley 

@mathr @paul @hecanjog Can we hear that anywhere?

Sign in to participate in the conversation
post.lurk.org

Welcome to post.lurk.org, an instance for discussions around cultural freedom, experimental, new media art, net and computational culture, and things like that.