NotebookLM says "transcript not available." Here's why, and the actual fix.

Published: October 24, 2026 (UTC)

If you've pasted a YouTube link into NotebookLM and gotten some version of "this video cannot be imported" or "transcript not available," here's what's most likely happening: NotebookLM reads YouTube's own caption track. If the video doesn't have captions, auto-generated or otherwise, there is nothing for it to read, and the import fails.

That's not a rare edge case. A meaningful share of YouTube content has no usable caption track: older uploads, videos where the creator turned captions off, non-English content where the auto-captions are poor, live-recorded lectures uploaded without post-processing, and plenty of small-channel content that was never captioned at all. If you're a student pulling a professor's lecture into your notes, or a researcher working through a channel's back catalog, you will hit this wall regularly.

Why this matters more than it sounds: the workaround most people reach for is transcribing the video by hand, which defeats the point of using an AI notebook tool. You wanted to skip the manual work, not do it before you're allowed to start.

The fix: transcribe the audio instead of reading a caption track that doesn't exist. I built ChannelScout as a research tool for my own work, and it handles this case. If a video has YouTube captions, it uses them (fast). If it doesn't, a single-video request falls back to Whisper AI, which transcribes the actual audio, so "no captions" stops being a dead end. You get a full, timestamped transcript either way, from the same paste-a-URL flow.

Once you have the transcript, getting it into NotebookLM (or any other AI tool) is one export: Markdown with clickable timestamp links, plain text (with or without timestamps), SRT, CSV or HTML. You're no longer depending on YouTube's captions, you're handing NotebookLM a text source it already knows how to read.

Concretely:

1. Paste the lecture or video URL into ChannelScout.
2. If it has no YouTube captions, it takes a bit longer, since audio transcription isn't instant (often a few minutes, longer for long videos), but you don't have to do anything differently.
3. Export as Markdown, or whatever format your workflow needs.
4. Add it to NotebookLM as a pasted or uploaded text source instead of a YouTube link.

The free plan includes 25 captioned transcripts a month plus 3 Whisper transcriptions (up to 60 minutes of audio in total), which is enough to try this on a real lecture before deciding whether you need more. Basic ($9/month) raises that to 10 Whisper transcriptions and 250 minutes, Pro ($19/month) to 25 and 550 minutes, and Studio ($39/month) to 50 and 1,000 minutes.

If you're working through a whole course or channel, channel jobs run in the background and are captions-only by default. On the plans that include it, you can switch a job to "Captions, then Whisper" so uncaptioned videos get transcribed too.

YouTube is not going to caption every video on the platform, so "transcript not available" is going to keep showing up. It doesn't have to be where your research stops.