Thursday, December 31, 2020

podcasting

Introduction

Many people are building their own podcasts at the moment. The intent of this project is not for others to be able to follow my podcast, but to preserve the podcasts that were most influential to me. It would be good to continue to listen to stuff that is currently being released, but have my own local server generate a podcast xml to be able to keep the "cleaned up" versions of audio that I will download.

Audio Processing

Testing out audacity. So far, looks like takes some amount of time to clean up to only the sermon segment of interest. Using copy & paste seems to work very well for picking portions of tracks

After it is clipped, then use the "truncate silence" function. If audio is high enough quality, then can use -50dB. Silence is tricky because it ending consonants can affect algorithm. Using -35dB and 250ms, was too short and caught sentences. Using 500ms was more clean. Guess is that consonants can dip into -35dB, then time-short silence threshold breaks understanding. Would like to do some analysis on how long do most people's words take in ms (eg. ms per syllable, then figure out what is the most efficient).

Note that one can also delete words that are unintended: um, uh, like, etc. but manual editing

Process

This is using the guides from here: dir2cast github   user1

Needed apache2 and php-xml from apt-get.

My final implementation:

 http://default-inspiron-3647/podcasts/dir2cast.php?dir=archive

Conclusion

Only problem is that original theory of hosting on off-internet server would be flouted by podcast app requiring their own curated link (see pca.st links). Thus, motivating me to move away entirely from "free" app that used to be $4.

I listen to podcasts at 1.8x and would like to preserve "trim silence" feature, going to add audacity scripts to "trim silence" before piping into any other podcast app. Thus will be good to process weekly sermons and preserve ones.

Very disappointing that there is no automated tool to generate local URL to pass directly to the app. I could process each file individually and save in google drive, sharing over to podcast app (profile > files)

No comments:

Post a Comment