This will listen to podcasts so you don't have to!
Install asdf
and docker
. Then run the following:
asdf install # install erlang/elixir
docker compose up -d # start db
mix setup # deps, db migrations, tailwind, esbuild
iex -S mix phx.server # start server in interpreter
In order to run whisper
(the model that converts audio to text) you need ffmpeg installed.
This is based on a small series of tutorials by Code and Stuff. These videos are linked below:
- Part 1: Transcribe Podcasts with Whisper AI & Elixir in Livebook
- Part 2: Boost Your AI Projects: Cloud GPUs + Elixir
In the tutorials he transcribed podcasts using code executed from a Livebook. The goal of this project is to create a website to perform AI-based transcription, with Phoenix, Ash, and Liveview.
alias Podcaster.Podcast
# Elixir Outlaws Podcast RSS URL
url = "https://feeds.fireside.fm/elixiroutlaws/rss"
show = Podcast.create_show_from_rss_feed_url!(url)
Podcast.create_episodes_from_show(show)
Podcast.update_transcripts(show)
To full clean and rebuild the project
rm -rf deps _build .elixir_ls
mix do deps.get + deps.compile + compile
- BUG: fix when multiple transcript requests are made
- UI: DARK MODE
- UI: landing page
- FEAT: run transcript logic in remote node
- FEAT: add summary and summary generation to episode
- FEAT: scheduled scanning for show updating