Hacker News Reader

this looks useful. i wonder how accurate it can detect chapters. i have been doing this work manually with audacity. and while the silence detector got me quite far, there always was some manual post processing to do because sometimes there were long silences within chapters or the chapter break was shorter than the others...

View on HN

Chapter detection is one thing that I'm working on. The speech detection thing is promising, there is a log of work to do for this. Especially non-english languages are difficult.

I'm also working on a online platform in my free time to share manually crafted chapter sets via api (like freedb), but I'm not sure that it will ever go online :-) I won't give it up either...

I have sort of an opposite wish for the chapter detection feature. I'm a big fan of podcasts that tell a story per season or whatever you want to call it. Lot's of them have an intro ad for an indeterminate period of time and signal the beginning of the episode with an intro blurb usually voice over of 'welcome to ..' with some background music. I really don't mind listening to the ads the first time through but would love to save whole seasons without the leader ads as an M4b file - I've thought about writing this in the past, just haven't gotten to it. The ability to detect those points in a series of mp3s and bundle them up into an m4b would be super useful in my book.

Nice idea... perhaps i will try to build this feature more generic and instead of only a chapter detection feature i will add a plugin interface, so that hooks could be defined for merge, split or chapter... But this is far far in the future ;)

Reply on news.ycombinator.com