10 ms·

Why not just speech to text?