Nice, but, why not speech to text solution?