Is anyone interested in having a discussion about NLP? The gist is my solution to a problem someone emailed me about the other day: how can I compare one audio clip to another and determine their similarity with regards to pronunciation and syntax, using Resin? My answer was that a syntactic comparison could be done with Resin if you first converted the audio clips to two pieces of text which you then compared to each other using the Resin search framework. Like I've done in the gist.
My question: what is state-of-the-art for comparing two audio clips syntactically?
Encoding pronunciation would mean encoding all words in their spoken form. Perfectly doable.
And do you have any feedback about how I performed the text comparison? Are you curious about the comparison technique?
[0] https://Github.com/kreeben/resin