Last verified: July 2026
NotebookLM Says the Video Language Isn’t Supported? Here’s Why
NotebookLM can only import what YouTube’s caption track gives it. For most non-English videos, that track is missing, auto-generated garbage, or simply not in a language NotebookLM can process — so the import fails no matter how popular the video is.
Failed to import source: this video’s language is not currently supported.
Phrasing varies by session — this is representative of what users report seeing.
Why this happens
NotebookLM doesn’t watch or listen to the video. It reads the caption track YouTube already generated — and that track is where non-English content falls apart.
Auto-captions never ran for that language
YouTube’s automatic captioning only covers a subset of languages well. Many regional and lower-resource languages get no auto-captions at all, so there is nothing for NotebookLM to read.
The captions exist but are unusable
Some languages get auto-captions that are heavily broken — wrong words, no punctuation, mixed-up characters. NotebookLM either rejects the track outright or imports something too garbled to summarize well.
The uploader captioned in a different language than the audio
A video spoken in Hindi or Arabic sometimes ships with an English caption track (or none at all), because the uploader only added captions in one language. NotebookLM sees a mismatch and bails.
Dialects and code-switching confuse the caption model
Videos that mix languages mid-sentence, or use a dialect YouTube’s captioning model wasn’t trained on, often produce a caption track too inconsistent for NotebookLM to treat as one supported language.
What happens with each tool
| Video situation | NotebookLM | ChatGPT | Gemini | YouTLDR |
|---|---|---|---|---|
| No captions at all | Import fails immediately | Can’t summarize | Falls back to metadata guesses | Transcribes the audio directly |
| Captions disabled by uploader | Import fails | Can’t summarize | Refuses or hallucinates | Transcribes the audio directly |
| Auto-captions only in another language | "Language not supported" error | Summarizes the wrong language, or fails | Mismatched or partial result | Transcribes in the video’s actual spoken language |
| Non-English video, low-resource language | Usually rejected | Inconsistent, often fails | Inconsistent, often fails | Supported — 125+ languages via audio transcription |
| 3-hour lecture or podcast | Import may time out or truncate | Can’t handle full length | Partial coverage | Supported up to 5 hours |
| Brand-new upload (captions not generated yet) | Import fails | Can’t summarize | Can’t summarize | Transcribes the audio directly, no wait needed |
How to actually get this video summarized
Paste the YouTube link into YouTLDR
Copy the video URL and drop it in — no caption track required, and no need to check which language YouTube thinks the video is in.
YouTLDR transcribes the audio directly
Instead of reading YouTube’s caption track, YouTLDR listens to the actual audio and transcribes it in the language it’s spoken in, across 125+ languages.
Get your transcript and TLDR with timestamps
Copy the transcript, export the summary, or keep asking follow-up questions in the studio — the same way you would have in NotebookLM.
Why YouTLDR works when the language is the problem
NotebookLM, ChatGPT, and Gemini all depend on YouTube’s caption track to know what was said in a video. If that track doesn’t exist for the video’s language, is disabled, or comes out garbled, there’s nothing for these tools to read — and the import fails regardless of how good the video is.
YouTLDR skips the caption track entirely. It transcribes the actual audio using a large-scale speech model, so it works the same way whether the video is in English, Hindi, Arabic, Japanese, Swahili, or any of 125+ supported languages — no auto-caption coverage required.
It also handles long-form content up to 5 hours, and you can try it on a video right now without creating an account. Paste a link, get a transcript and summary, and see for yourself before you decide whether to sign up.
Frequently asked questions
Is this NotebookLM’s fault?
Not exactly. NotebookLM is built to read source documents, including YouTube caption tracks — it was never designed to listen to audio. The limitation is architectural: no caption track in a supported language means no import, by design.
Will YouTube add captions for this video later?
Sometimes, for popular videos in well-supported languages. But many lower-resource languages never get reliable auto-captions, and uploader-added captions depend entirely on the uploader bothering to add them.
Does this work for live streams?
Only after the stream has ended and is available as a regular video. Live streams in progress can’t be transcribed yet.
How long can the video be?
YouTLDR supports videos up to 5 hours long, which covers most lectures, podcasts, and long-form interviews that would otherwise time out in caption-dependent tools.
What languages are supported?
YouTLDR transcribes audio in 125+ languages, including many regional and lower-resource languages that YouTube’s auto-captioning doesn’t cover well.
Do I need an account to try it?
No. You can paste a link and get a transcript and summary without signing up. Creating an account just lets you save your work.
Related fixes
NotebookLM Says the Video Language Isn’t Supported? Here’s Why
NotebookLM can only import what YouTube’s caption track gives it. For most non-English videos, that track is missing, auto-generated garbage, or simply not in a language NotebookLM can process — so the import fails no matter how popular the video is.
Try YouTLDR free