I am looking for an algorithm to compare two sound files by means of their phonetics. I want to evaluate some kind of score (e.g. 0: no similarity; 100: the same sound). Therefore, I loaded a computer generated and a recorded human voice saying “Hello” into Audacity (I know, it’s not that professional…) and looked at the waves.
https://dl.dropbox.com/s/p8o8tpsayo9xr14/hello.png (Unfortunately, I am not allowed to embed images)
Although both sound completey the same, the waves don’t really match that much.
Is there a way to detect the similarity between these two waves and calculate the mentioned score?
Here is an open source project that you may be interested in free speech recognition tools and applications. Here’s another one which is an open source toolkit for speech recognition. I’m sure if you “duckduckgo” you will find others if these are not to your taste. CHEERS!