Research: Can we develop on, or, contribute to

[tikikun]

why this matters? what we want

• We want to keep using semantics capture to have clear instruciton voice
• We want extra information from just the voice (highn low pitch etc)

https://github.com/zhenye234/xcodec

[tikikun]

open in the future

[tikikun]

We replace this with synthetic semantic tokens