3 ms·so then the training data is text, not audio?by hackernewds 3y agoso then the training data is text, not audio?