Pronounce
9, Apr, 2015Pronounce is a tiny rule based grapheme to phoneme (G2P) converter.
It achieves roughly 75% syllable accuracy on CMU pronouncing dictionary data. This is far lower than state of the art systems, but it has the advantage of being much smaller and simpler. In fact, the conversion takes place in a single self-contained function.
I wrote this for situations like sentiment analysis, or a speech input system where the user chooses from a list of options. These cases require less accuracy than a full speech to text system.
