google's speech recognition algo is cloud only. yes, you can run speech recogntion on your own local machine, but the output will not be nearly as accurate as google's cloud algorithm that has strengthened itself from countless iterations on a massive corpus. only way your local one would be good would be is if you trained it manually on owen, which would take quite a bit of work i'd imagine.
but yeah anyway, you can use google's cloud services and get outputed text. i got the transcripts from the streams listed above. feel free to do it on the others. i wouldnt post them publicly though, dont want to make it easy for the journalists to take things out of context just by searching for keywords. i guess you could send me the .txt at ursadelta at gmail and i'll keep track of them or something. just getting the full .txt transcript of the episode is fine
oh also there's an option in google cloud services to add timestamps when doing the text output. May as well check that option, the timestamps can be useful yet easily removed with regex if needed