How to transcribe and summarize a YouTube video

On a computer, open the video, expand the description and click “Show transcript”. A panel opens beside the player with the full caption text; clicking a line jumps the video to that moment, which makes quote-checking easy later. Select the text and copy it. Only videos with captions show the button, so if it's missing, the video may have no captions at all — skip to step 5.

Auto-generated captions have no punctuation in places and no speaker names, so a two-person interview can turn into a wall of unattributed text. If the video has several speakers, names or technical jargon, the free route will cost you more editing time than it saves — go to step 5 and transcribe the audio itself with a tool such as Notta, which labels speakers and handles long videos (up to 5 hours) in one pass.

Paste the transcript into Gemini or another assistant with a prompt such as “Summarize in five bullets with timestamps; list any figures quoted.” Or ask Gemini about the video directly (signed in, with Keep Activity on). On the watch page, Ask YouTube answers questions about select videos in select regions, though Google notes it “may sometimes produce inaccurate information”.

Add each public video's URL as a source. NotebookLM imports the caption text only, so it needs captions, and it may not import videos uploaded in the last 72 hours. It's a good free option for research across many captioned talks; for videos without captions, you'll still need a transcription service.

Paste the link into Notta Web (“Upload & transcribe”: public videos and Shorts, up to 5 hours) or HappyScribe (“Public link”; first 10 minutes free). Both transcribe the audio itself, so they work even when a video has no captions, and both label speakers — something YouTube's captions never do. Notta is the stronger option for longer videos: its 5-hour limit covers lectures, conference talks and long interviews, and its Pro plan (1,800 minutes a month) costs less than HappyScribe's Basic plan (120 minutes a month). Set the language to the video's spoken language before you start. On a Mac, MacWhisper Pro also transcribes YouTube links with on-device processing.

Notta offers an AI summary alongside the transcript, so you can go from link to summary without copying text between apps — a real time-saver when you're processing several videos. HappyScribe also generates summaries in-tool. Check which plan includes AI summaries before you rely on this route.

Jump to the timestamp and listen before quoting a name, number or claim. Auto-captions in particular can garble proper nouns and figures. For a fuller method, see our guide to verifying quotes in AI transcripts.