Listen to a YouTube video like a podcast
Some videos are worth an hour of your attention. Most are worth twelve minutes of listening while you do something else. Paste a link, choose what you want out of the video, and have it read to you in a natural voice — or keep it as an MP3 and listen later, offline, at whatever speed suits you.
Free · no sign-up · unlimited
How it works
Paste the video link
Any youtube.com or youtu.be address. The transcript loads straight away, free and without an account.
Choose what to listen to
A short summary for the gist, key insights for the arguments, or full notes if you are studying. Each produces a different length of audio.
Press listen, or keep the MP3
The text is read aloud in a Google Neural2 voice. Download it and it behaves like any other podcast episode on your phone.
What you are actually listening to
This is worth being precise about, because it is the difference between this and a downloader. You are not listening to the video's own audio. You are listening to text that was written from the video's transcript — a summary, a set of insights, a page of notes — and then spoken by a synthetic voice.
That has a real consequence: the audio is shorter than the video, usually by a lot. A fifty-minute conference talk becomes six or seven minutes of narration. Nothing is lost that the summary kept, and a great deal is lost that it did not. If you need every word, read the transcript instead; it is free and it is complete.
It also means the audio has no music, no crosstalk, no laughter and no pauses for effect. For a lecture or an interview that is a straight improvement. For a performance it is not, and you should watch the video.
Why listening beats watching for some material
Video demands your eyes. That is fine when there is something to look at — a demo, a diagram, a face — and wasteful when there is not. A very large share of long-form YouTube is a person talking, and for that material the picture carries almost nothing the words do not.
Audio also survives being interrupted in a way video does not. You can walk away from a podcast and pick it up; walking away from a video means finding your place again in a timeline that all looks the same.
And listening happens in time you already have. Commutes, cooking, walking, the gym. The realistic alternative to a twelve-minute narration is not watching the fifty-minute video — it is never getting to it at all.
Choosing the right length
The tool you pick decides how long you listen for. It is worth matching that to why you opened the video.
| What you want | Pick | Roughly |
|---|---|---|
| Is this worth my time at all | Micro summary | Under a minute |
| The gist, to talk about it | Short summary | Two to three minutes |
| The argument and its support | Key insights | Four to six minutes |
| To learn it properly | Outline or Cornell notes | Ten minutes or more |
Voices, speed and languages
Narration uses Google's Neural2 voices where they exist for a language, and WaveNet where they do not — Turkish, for instance, has no Neural2 voice yet. Both are a long way from the flat robotic reading people associate with text-to-speech, but they are not indistinguishable from a person, and long stretches make that clearer.
You can have the output in a different language from the video. The tools translate as they generate, so a Spanish talk can produce English notes read in an English voice. That is two lossy steps rather than one, and it is worth checking anything you intend to quote against the original transcript.
Playback speed is a matter for whatever you listen in. Downloading the MP3 hands the file to a real podcast player, which is better at 1.5× and at remembering where you stopped than any web page.
Getting it onto your phone
The MP3 is a plain audio file with no restrictions, so anything that plays audio will play it. Most people either save it to their phone directly or drop it into whatever podcast app they use, most of which accept local files.
Because the file is generated from your own choice of tool and language, two people listening to the same video can end up with quite different episodes. That is the intended behaviour: the video is the source, not the product.
What this can't do
- The narration is of AI-generated text, not the video's own audio. It is a great deal shorter, and anything the summary dropped is gone.
- Synthetic voices are good, not human. Over ten minutes or more, that becomes noticeable.
- Videos with no caption track produce nothing to summarise or narrate.
- Narration and the AI tools are part of the subscription. Reading and searching transcripts is free and unlimited.
Frequently asked questions
Is this the video's actual audio?
- No. It is a summary or set of notes generated from the transcript, then read aloud. It is much shorter than the video and does not contain the original speakers' voices.
Can I download it?
- Yes, as an MP3. It is an ordinary audio file and plays anywhere.
How long is the audio?
- It follows the tool you pick — from under a minute for a micro summary to ten minutes or more for full notes, largely regardless of how long the video was.
Does it work for videos in other languages?
- Yes, and the output can be in a different language from the video. Translation and summarising are both lossy, so check anything you plan to quote.
Is it free?
- Reading and searching the transcript is free, unlimited and needs no account. Narration and the AI tools are part of the subscription.
Will it work on a music video?
- Rarely. Music videos usually have no usable caption track, and there is nothing to summarise if there is no speech.