yt-dlp Transcript Workflow

The Problem

Sites like Rumble don’t auto-generate captions the way YouTube does. For long livestreams, you often only need a transcript of a specific segment.

Segment + Transcribe Workflow

# 1. Download only the portion you need (audio only)
yt-dlp --download-sections "*1:30:00-2:00:00" -x --audio-format mp3 "VIDEO_URL" -o segment.mp3

# 2. Transcribe with Whisper
whisper segment.mp3 --model medium --language en

Notes

--download-sections takes *START-END format (e.g., *10:00-25:00 or *1:30:00-2:00:00)
Scrub through the video in-browser first to find the timestamps you want
Whisper runs well on GPU (RunPod) for long segments
Check if subtitles already exist before transcribing: yt-dlp --list-subs "URL"

Full video transcript (no segmenting)

# YouTube — grab existing auto-captions
yt-dlp --write-auto-sub --sub-lang en --skip-download "YOUTUBE_URL"

# Any site — download audio and transcribe
yt-dlp -x --audio-format mp3 "VIDEO_URL" -o audio.mp3
whisper audio.mp3 --model medium --language en

Notable Sites Supported by yt-dlp

Video Platforms: YouTube, Vimeo, Dailymotion, Rumble, Bilibili, PeerTube

Social Media: Twitter/X, Facebook, Instagram, TikTok, Reddit, LinkedIn

Live Streaming: Twitch (live + VODs + clips), Kick

Audio / Music: SoundCloud, Bandcamp, Mixcloud, Audiomack

News / Media: BBC, CNN, NBC, CBS, ABC, NPR, ARD/ZDF, France TV, NHK

Education: TED, Khan Academy, Coursera, Udemy

Podcasts: Spotify (podcasts only), Apple Podcasts

Other: Internet Archive, Google Drive, Dropbox, Imgur, Niconico