I have a collection of podcasts, in particular the "Talkingmachines" Podcasts ("Human conversation about machine learning"), see http://www.thetalkingmachines.com/blog/

There are 22 episodes, each 1 hour long. MP3 files are available on my local disk.

Now when I want to search the audio files for the episode where they talk about, say, the kmeans++ algorithm , what should I do?

Specifically, I want to search for that episode where a male voice (Ryan) talks about kmeans++ in the first half of the podcast.

Again, what should I do? Is there a desktop tool available that can do this search? Should I upload the files to some webservice that can do audio recognition?

| improve this question | | | | |

Transcribe then topic model the recordings. This will let you know which podcast talks about the subject of interest, then you can search the transcript. If you really need it to pinpoint the moment when the subject was discussed, you'll need a temporal topic model like so but that would add a good bit of complexity.

Google has a good service for the first part. You can use an open source library like gensim for the second part.

| improve this answer | | | | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.