No account · No extension · Free SRT
Free YouTube transcript, with or without captions
This reads the audio rather than YouTube's caption track, so it works on videos where the transcript panel and every browser extension come back empty.
- Link
- Transcribe
- Transcript
Watch links, youtu.be and Shorts. Up to 20 minutes.
Set it yourself for short clips, heavy accents, or mixed languages.
English translates as it transcribes, in one pass. It's the only target.
Free, no account. Up to 20 min per file, 2 per hour.
Start here
If the video has captions, YouTube is faster
Open the video, click the three dots under it, choose Show transcript. If that works and you only need to read it, you're done — close this tab, no tool required.
Worth saying first, because most pages competing for this search quietly hope you've forgotten it exists. The same goes for the extensions — they read that same caption track, which is why they're instant and why they fail in exactly the same places YouTube does.
This tool takes a different route. It downloads the audio and transcribes it with Whisper large-v3, which is slower and capped at 20 minutes — and which works in four situations where reading the caption track gets you nothing:
| Situation | YouTube's panel & extensions | Here |
|---|---|---|
| Captions disabled by the creator | Nothing to read | Works — reads the audio |
| Language YouTube doesn't auto-caption | Nothing generated | Around 100 languages |
| Auto-captions are visibly wrong | Copies the same mistakes | Independent second pass |
| You need a subtitle file | No download for viewers | SRT or VTT, free |
| Video is over 20 minutes | No length limit | Rejected — see below |
That last row is a real loss, not a hedge. On a long video that already has captions, YouTube's panel wins outright and this page is the wrong tool. Saying so is cheaper than letting you find out at an error message — and it's the same reasoning behind what "free" usually means in this category.
When the auto-captions are wrong
A second opinion, not a verdict
Auto-captions mangle names, jargon and accented speech, and they get worse with music under the voice. Reading them from a different tool gets you the same mistakes.
Every browser extension and free transcript site is reading the same caption track YouTube already generated. If that track is wrong, all of them are wrong in identical ways — there is nothing in that pipeline that could disagree.
This runs Whisper large-v3over the audio independently, so where the two disagree you have something to compare. Neither is authoritative and we're not going to claim ours is: two independent passes on a difficult name is genuinely more useful than one confident number from either side.
What actually decides accuracy is the recording, not the tool. A studio podcast transcribes near-perfectly on both. A handheld phone in a busy room is hard for everything.
How it works
Paste, wait, export
01
Paste the link
Watch links, youtu.be and Shorts all work. Nothing downloads to your device.
02
Set the language
Or leave it on auto-detect. Set it for short clips and mixed-language videos.
03
Read or export
Copy the text, or download TXT, SRT or VTT — no account at any point.
Long videos
Past 20 minutes, there's a longer route
20 minutes covers most interviews, lectures and music videos. It does not cover a two-hour podcast, and pretending otherwise would just waste your time — a longer link is rejected rather than half-transcribed.
If the video has captions, stop here and use YouTube's own transcript panel; it has no length limit. If it doesn't, the workaround runs through three other tools here and takes a few minutes: pull the audio out of the video, split it at natural pauses so no sentence is cut in half, then transcribe each section. Timestamps restart at zero in each piece, so joining plain text is easy while building one continuous caption file means adding each section's offset.
01
YouTube to WAVPull the audio out of the video.
02
Silence SplitterCut it at natural pauses, not fixed times.
03
Audio to TextTranscribe each section in turn.
Honest limits
What this won't do
Worth knowing before you paste a link rather than after.
- Slower than reading the caption track
- Transcribing audio takes up to about a minute; an extension that reads existing captions is instant. That's the trade for working when there are no captions to read.
- No playlists or channels
- One video per run. There's no bulk mode.
- No speaker labels
- An interview comes back as continuous text, not "Host / Guest".
- Private and region-blocked videos won't load
- If the server can't reach the video, there's no audio to work from.
- No playback here
- Unlike the file-based tools, there's no audio to click a transcript line against — watch the video alongside it instead.
For what actually degrades a transcript and when to set the language manually, read the transcription accuracy guide.
Next
More free tools
Audio to Text
Transcribe audio to text with timestamps, free and without an account.
Video to Text
Upload a video and get a transcript or subtitle file.
YouTube to WAV
Convert YouTube videos to lossless WAV audio for sampling and DJ sets.
Silence Splitter
Split one long recording into separate tracks at silent gaps.
Questions
Frequently asked questions
Copyright & fair use
You're responsible for how you use a transcript of someone else's video. Personal reference, accessibility, study and quotation are generally reasonable; republishing a transcript as your own content is not. AudioForges doesn't host or redistribute video, audio, or the transcripts produced here.
Published · limits and model re-checked . Whisper large-v3 · 20 minutes per video · 2 per hour.