Ask HN: Are there human sound generation emulators?

[dmingoddd]

Like we have 3D programs that can trace light and create images. Are there any acoustic generators that mimic how a human produces sound using software and then give the output. To be clear this is the opposite of recording sound and playing it back. It is basically defining the environment where the sound would get created, have the sound waves mocked by software and play the actual sound that is produced. If a collision happens in a tin room how would the sound waves travel and what sound would they produce.

One of the things I can think of is this would go a long way towards generating human like speech, especially if breathing limitations, internal organs, shape of the mouth and tongue are mocked correctly.

[tomcam]

It's called modeling the human vocal tract. Sampling, then tweaking with DSP has proven to be a better approach.

[dmingoddd]

But like a rendering engine are there sound software that can generate and simulate realistic sounds without being fed recorded stuff in.....

[tomcam]

Yes, I get what you want. This approach is used in many modern synthesizers, such as the Korg Kronos. It's just that approach proved to be too complicated to get working well for speech to text. I was a little surprised myself. Google for vocal tract modeling and you'll see it pretty much died out a decade or so ago.