Comment by simonw

That was about the text-to-speech model, the speech-to-text one was release in January.