How accurate is the transcription?

It depends on the model you pick — accuracy scales with model size, and that is physics, not marketing. Fast is a quick draft that makes real mistakes; Accurate (the default) handles clear speech well; Best is noticeably stronger, especially for non-English audio, accents and noise. Cloud services running giant models are still somewhat more accurate — the trade-off here is total privacy and zero cost. For audio that matters, use Best and skim the result.

Which languages are supported?

All 29 languages in the selector, each tested with real recorded speech, not synthetic voices. In our detection test the original 27 languages were identified correctly 27 of 27 times; Urdu is not detected automatically, so choose it in the list. By default the language is detected automatically (we ask the model which language it heard before transcribing) and the transcript comes out in that same language. Each language also has its own page, which preselects it and, where our tests showed the smaller models struggling (Danish, Finnish, Romanian, Greek, Turkish, Hebrew, Vietnamese, Thai, Hindi), opens with the most accurate model already chosen.

Can it translate the audio to English?

Yes, into English and into the other languages listed in the Output menu. Set Output to "Translate to English" and Whisper translates natively while transcribing. For any other target language, choose it in the Output list: the text is transcribed first and then translated on your device by an open-source translation model (about 100 MB, downloaded once), still without uploading anything.

Can it transcribe a YouTube or TikTok video?

Not from a link — browsers can't access other sites' video streams, and platform terms prohibit it. It works with video files on your device: recordings, downloads you own, screen captures, lectures.

Is there a length or file-size limit?

There is no per-minute charge, but this is not an unlimited-capacity service. Files over 8GB are rejected, and smaller files can still exceed your device's memory. Start with a short recording; trim or split long files, especially on a phone.

Is my video really private?

In the default Private mode, yes: audio is extracted and transcribed entirely on your device, and nothing — not the video, not the audio, not the transcript — is ever sent to us or anyone else. There is also an optional Instant mode (Chrome only) that skips the model download by using the browser’s built-in speech service; that one sends the audio to Google’s servers, which is why it is clearly labelled and never the default.

What is Instant mode?

A no-download option for Chrome users. Instead of downloading a speech model, it hands the audio to the browser’s built-in speech service. Trade-offs, stated plainly: the audio is sent to Google, it runs in real time (a 10-minute file takes about 10 minutes), timestamps are approximate, and translation is not available in this mode. Pick it when you want speed and don’t mind the trade; stay in Private mode when the recording is sensitive.

Video to Text — Transcribe Free, No Upload, No Signup

Turn video or audio into editable text for free, with no signup. In Private mode (the default) everything runs on your device and nothing is uploaded. Review, copy or download TXT.

Model

Spoken

Output

Timestamps [0:05]

🌍 The spoken language is detected automatically and the transcript comes out in that same language. Set it manually above if your audio is hard to hear, and use Output to translate into English.

⬇️ One-time download: 104 MB (Accurate model 73 MB + 31 MB audio engine, shared with the video tools) — only the first time: it stays saved in your browser, so next time it starts right away. (Translating or switching model needs its own file; a browser can clear this cache if storage runs very low.) Processing takes about 4-8 min per 10 minutes of audio on a typical laptop.

What exactly is downloaded — and is it safe?

• The speech model is OpenAI's Whisper, open source, converted to run in browsers. The audio engine is FFmpeg, the open-source tool behind most video software.
• They come from Hugging Face and unpkg — public repositories used by researchers and developers worldwide, not from us.
• They are not programs. Nothing is installed, no permission is asked, and your browser keeps them in its own cache exactly like it keeps a website's images.
• They can't read your files or touch anything outside this page. Remove them any time by clearing this site's data in your browser.
• Your audio never goes the other way: this site has no server receiving files.

Drop your video or audio here or click to choose

transcribed on your device by OpenAI's Whisper — your recording is not uploaded

ℹ️ Use a recording on your device that you own or have permission to process. This tool cannot fetch YouTube or social video links. Start with a short clip. Larger models can improve accuracy but require more memory and download data. Processing time varies by device, model and recording length. Always review the result.

Your first result

From a recording to text you can copy and review

1. Start with a short clipChoose a clear recording you own or have permission to use. Select the spoken language if you know it.
2. Stay in Private modeYour device does the work. Processing files download first; keep this tab open. Fast uses a smaller model, with an accuracy trade-off.
3. Review, then exportCheck names, numbers and punctuation, then copy the transcript or download TXT. This is a draft transcript, not an automatic summary.

No recording handy? Use our short English sample.

Download the synthetic test recording, choose English above, and select the downloaded file. The sample and reference output contain no personal recording. Your result may vary by model.

Download sample audio See the reference transcript

First-run downloads and processing take time. Long recordings may exceed your device's memory; start with a short clip, or trim it first. Private mode keeps your recording on your device. Optional Instant mode uses the browser's external speech service.

Making captions for your own video? Open Subtitle Generator.

Turn a lecture, interview, voice memo or your own video into a draft transcript you can review and reuse. In the default Private mode, OpenAI's Whisper runs in your browser and your recording is not uploaded. Copy the text into your notes or download TXT; subtitle exports are also available.

Choose Fast for a smaller model, Accurate for a more demanding option, or Best when you can allow more memory and processing time. The first use downloads the audio engine and speech model. Files are cached when possible, but browser cleanup, another model or translation may require further downloads. There is no per-minute charge. Speed and accuracy depend on your device and recording; start with a short, clear clip and review names, numbers and important statements.

More sizes & options

Frequently asked questions

How accurate is the transcription?: It depends on the model you pick — accuracy scales with model size, and that is physics, not marketing. Fast is a quick draft that makes real mistakes; Accurate (the default) handles clear speech well; Best is noticeably stronger, especially for non-English audio, accents and noise. Cloud services running giant models are still somewhat more accurate — the trade-off here is total privacy and zero cost. For audio that matters, use Best and skim the result.
Which languages are supported?: All 29 languages in the selector, each tested with real recorded speech, not synthetic voices. In our detection test the original 27 languages were identified correctly 27 of 27 times; Urdu is not detected automatically, so choose it in the list. By default the language is detected automatically (we ask the model which language it heard before transcribing) and the transcript comes out in that same language. Each language also has its own page, which preselects it and, where our tests showed the smaller models struggling (Danish, Finnish, Romanian, Greek, Turkish, Hebrew, Vietnamese, Thai, Hindi), opens with the most accurate model already chosen.
Can it translate the audio to English?: Yes, into English and into the other languages listed in the Output menu. Set Output to "Translate to English" and Whisper translates natively while transcribing. For any other target language, choose it in the Output list: the text is transcribed first and then translated on your device by an open-source translation model (about 100 MB, downloaded once), still without uploading anything.
Can it transcribe a YouTube or TikTok video?: Not from a link — browsers can't access other sites' video streams, and platform terms prohibit it. It works with video files on your device: recordings, downloads you own, screen captures, lectures.
Is there a length or file-size limit?: There is no per-minute charge, but this is not an unlimited-capacity service. Files over 8GB are rejected, and smaller files can still exceed your device's memory. Start with a short recording; trim or split long files, especially on a phone.
Is my video really private?: In the default Private mode, yes: audio is extracted and transcribed entirely on your device, and nothing — not the video, not the audio, not the transcript — is ever sent to us or anyone else. There is also an optional Instant mode (Chrome only) that skips the model download by using the browser’s built-in speech service; that one sends the audio to Google’s servers, which is why it is clearly labelled and never the default.
What is Instant mode?: A no-download option for Chrome users. Instead of downloading a speech model, it hands the audio to the browser’s built-in speech service. Trade-offs, stated plainly: the audio is sent to Google, it runs in real time (a 10-minute file takes about 10 minutes), timestamps are approximate, and translation is not available in this mode. Pick it when you want speed and don’t mind the trade; stay in Private mode when the recording is sensitive.

Guides

Related tools

🔒 Private mode (the default): your recording never leaves your device. Instant mode is optional, you switch it on yourself, and it sends the audio to the browser's speech service.