Introducing Whispertalk: Whisper for Predicting Subsequent Utterances with Audio Input

[hojinYang]

Hi all, I wanted to share Whispertalk. By fine-tuning the Whisper, we've made it to predict subsequent utterances based solely on audio input, optionally supplemented with prompts or contextual information. You can find the demo, Colab, and technical report in our repository here.

If you are interested in our project, have feedback, or would like to collaborate on further research, we would greatly appreciate hearing from you. :) Thank you!