Guide
How to get transcripts of Lex Fridman Podcast videos
Three-hour conversations are impossible to skim. Transcripts let you find the exact exchange you half-remember.
Get every available transcript in three steps
- Paste the podcast link. BulkTranscripts detects whether it is a video, playlist, or channel automatically.
- Press Get transcripts. Videos are processed in parallel with live progress, and anything already in your library is served instantly from cache.
- Export. Download a ZIP with one file per video, or a single combined document in TXT, Markdown, SRT, VTT, CSV, JSON, or AI-ready format.
What people do with Lex Fridman Podcast transcripts
The common use is to pull quotes for writing, or load a guest's full appearance into an AI to summarise their argument. Because the combined export carries each video's title, channel, date, and link in a header, an AI model can cite which video an answer came from instead of blending everything together.
About the Lex Fridman Podcast archive
The Lex Fridman Podcast is several hundred interviews, many over three hours, with scientists, engineers, founders and public figures. Episodes are numbered and titled by guest, which makes the archive easy to filter by person, and the episodes we have pulled all carry captions uploaded by the channel, so names and technical terms come through as spoken. A single guest can have multiple appearances years apart.
Questions the full transcript set can answer
Load the combined export into NotebookLM, ChatGPT or Claude and ask:
- Across all of a guest's appearances, how did their view on AI risk change?
- Which episodes mention a specific book, and what was said about it?
- Pull every quote about consciousness with the guest name and episode number.
What Lex Fridman Podcast looks like as text
Measured from Lex Fridman Podcast videos our users have already transcribed, pulled from our production library on September 27, 2026. It is a sample of the archive, not the whole thing.
| Measure | Result | What it means |
|---|---|---|
| Videos already transcribed | 12 | Upload dates up to September 2026 |
| Hours of audio | 40 | Average 198 min, longest 315 min |
| Words per video | 31,945 on average | About 64 printed pages each |
| Speaking pace | 162 words a minute | Useful for estimating an episode you have not pulled yet |
| Caption source | 12 channel-uploaded, 0 auto-generated | Channel captions are preferred when both exist |
For sizing: 100 episodes at this length is roughly 3.2 million words and costs 100 credits, one per video. That is more than most AI chats will hold at once, so use the combined export in NotebookLM, or connect the MCP server and let the assistant fetch only the episodes a question needs.
Questions
Are Lex Fridman Podcast transcripts auto-generated?
None of the 12 videos from Lex Fridman Podcast in our library needed auto-captions: all of them carry captions uploaded by the channel, so names and numbers come through as spoken.
Is there a limit on how many Lex Fridman Podcast videos I can transcribe?
You can run a whole channel or playlist in one job. Large archives are processed in batches with live progress, and everything you have already pulled is cached so re-runs are instant.
Do I need to sign in?
No. Paste a link and go.
What if a video has no captions?
It is skipped and reported in the results, and the rest of the job continues. Videos with creator-written captions are preferred over auto-generated ones when both exist.
Can I get one file instead of hundreds?
Yes. Choose the combined export to get the entire channel or playlist as a single document, which is what you want for NotebookLM, ChatGPT, or Claude.