Batch podcast RSS ingestion #13

New issue

Closed

opened 2026-09-01 20:40:24 +00:00 by Mark · 0 comments

Mark commented

2026-09-01 20:40:24 +00:00

Owner

Lex Fridman feed returned 502 episodes in yt-dlp --dump-json. Currently you'd run the CLI once per episode. Want a batch mode:

transcribe --feed https://lexfridman.com/feed/podcast/ --episodes 1-10 --output-dir /tmp/lex/

What needs building:

--feed URL mode that enumerates the RSS feed
--episodes N or --episodes 1-10 selector (defaults to latest 1)
--output-dir with auto-named files ({date}_{title}.json)
Parallelism: --parallel N to transcribe N episodes concurrently (limited by VRAM — base model = 1 at a time on 2060)
Per-episode error isolation: one failed episode doesn't kill the batch
Resume: --skip-existing so re-runs pick up where they left off

Why: podcasters who want to make their episodes searchable will want the whole feed, not one URL at a time.

Lex Fridman feed returned 502 episodes in `yt-dlp --dump-json`. Currently you'd run the CLI once per episode. Want a batch mode: `transcribe --feed https://lexfridman.com/feed/podcast/ --episodes 1-10 --output-dir /tmp/lex/` **What needs building:** - `--feed URL` mode that enumerates the RSS feed - `--episodes N` or `--episodes 1-10` selector (defaults to latest 1) - `--output-dir` with auto-named files (`{date}_{title}.json`) - Parallelism: `--parallel N` to transcribe N episodes concurrently (limited by VRAM — base model = 1 at a time on 2060) - Per-episode error isolation: one failed episode doesn't kill the batch - Resume: `--skip-existing` so re-runs pick up where they left off **Why:** podcasters who want to make their episodes searchable will want the whole feed, not one URL at a time.