I’m using the built-in speech recognition block for my AI chat app. Configuring APIs for things like this can be complicated but it should be possible. For me, the Thunkable speech recognition works pretty well so I didn’t end up using an API for that.
I do use a Google Cloud Voice API for text-to-speech because of the far superior sound of the voice and more natural pacing of speech when compared to Thunkable’s text-to-speech block. You might look at this: https://cloud.google.com/speech-to-text.