Hi,
I’ve just seen that NASA has releases their multimedia db in the public domain. It seems they also have audios with text transcriptions like this one.
Wondering how difficult would be to scrap all audios and texts to match and use with deep speech.