Talking avatar JS sample - kindly read the project details first.
Budget: $10 – $30 USD
We developed a game-engine based talking avatars for a cinematic multi characters project. I was checking the potentials of doing a light version online and saw potentials in an old JS based platform (will share links below and more useful content once we discuss the project with you). So, we want to try that in a small "project" that requires no sophisticated stuff before making decisions. If the stuff below were done 6 years ago, I imagine it is more doable now.
Follow the links to make the point clear:
Here is a talking avatar with saved wav files you can click to test (open in chrome, and give it time to load for all the links below):
https://webglstudio.org/users/gerard/lipsyncE4/
Now, instead of saved wave files, we want to be able to edit that sample, as an example, so we can post wave files to it (maybe via a JS api or something that we can post the wave files through).
Now check this one:
https://webglstudio.org/users/gerard/eliza/
Here is the code github for that project:
https://github.com/gerardllorach/eliza
The speech to text and text to speech functions there are very good. If you can use them, the end project will be as following:
- The first link avatar (or another good one, we can share links to many).
- The listening and speech to text features used in the second link. Once the user "talks" to the avatar, the script convert the speech to text.
- Your project has the api (or equivalent) to post that text to an external system (we have already), and get a response text from that system. (we have another alternative adding a python wiki script that takes the text and respond with an answer for testing the sample task. But we will also need a more versatile approach - like via api- to communicate with other platforms)
- Your script convert the responded text to wav and play it with lipsync (as in the first link)
I believe most, if not all the code - beyond the api- for that is available in the demos on that platform. You can play around it here:
https://webglstudio.org/users/
Note that this was all built 6 years ago. If you have something better, we don't mind it. We want a development long-term partner, not a coding freelancer. As long as we do not turn this sample trial to a mega project. That will happen if the sample provided satisfactory results.
Follow the links to make the point clear:
Here is a talking avatar with saved wav files you can click to test (open in chrome, and give it time to load for all the links below):
https://webglstudio.org/users/gerard/lipsyncE4/
Now, instead of saved wave files, we want to be able to edit that sample, as an example, so we can post wave files to it (maybe via a JS api or something that we can post the wave files through).
Now check this one:
https://webglstudio.org/users/gerard/eliza/
Here is the code github for that project:
https://github.com/gerardllorach/eliza
The speech to text and text to speech functions there are very good. If you can use them, the end project will be as following:
- The first link avatar (or another good one, we can share links to many).
- The listening and speech to text features used in the second link. Once the user "talks" to the avatar, the script convert the speech to text.
- Your project has the api (or equivalent) to post that text to an external system (we have already), and get a response text from that system. (we have another alternative adding a python wiki script that takes the text and respond with an answer for testing the sample task. But we will also need a more versatile approach - like via api- to communicate with other platforms)
- Your script convert the responded text to wav and play it with lipsync (as in the first link)
I believe most, if not all the code - beyond the api- for that is available in the demos on that platform. You can play around it here:
https://webglstudio.org/users/
Note that this was all built 6 years ago. If you have something better, we don't mind it. We want a development long-term partner, not a coding freelancer. As long as we do not turn this sample trial to a mega project. That will happen if the sample provided satisfactory results.