Run watson Text to Speech output through the Speech to Text service to get word timings, then bundle the two together - View it on GitHub