Speech into subtitles, on your own machine
Turn a recording into text and timed subtitles in this browser, edit it against the playback, and export SRT, VTT, TXT or JSON. The speech model runs here — your media, your transcript and your edits are never uploaded.
Nothing here is uploaded. Transcription runs on this device using an on-device model. The model files are fetched once from a CDN and cached by your browser; the media, the transcript and your edits never leave it.
The recording
A file from this device, or the audio of another tab.
Drop a video or audio file, or click to choose
MP4, WebM, MOV, MP3, WAV, M4A… processed locallyTranscribe a video playing in another browser tab — a lecture, webinar, or any web video — without downloading anything. Your browser records the tab's audio locally, in real time, then transcribes it on your device. In the picker, choose the tab and tick "Also share tab audio".
Language and model
Balanced suits most recordings. Fast is for a phone; Accurate is worth the wait on a hard one.
Free video to text, video to SRT — without uploading your file
This is a free video transcriber and SRT generator that runs entirely in your web browser. Drop in a recording and it converts speech to text on your own device using a local AI speech-recognition model (Whisper), then produces timestamped subtitles you can edit and export as SRT, WebVTT, plain text, or JSON. Because the audio is processed locally, your file isn't sent to Toolsfully or any transcription service.
Why local, in-browser transcription matters
Keeping the work on your device is genuinely useful when a recording is sensitive or simply large:
- Interviews and meetings — transcribe conversations without handing the audio to a third party.
- Lectures and podcasts — turn long recordings into searchable text and captions.
- Social-media videos — generate SRT/VTT captions to boost reach and accessibility.
- Personal and confidential recordings — medical, legal, or private notes stay on your machine.
A quick, honest note on privacy: the transcription itself happens locally and your media is not uploaded. Like any website, the page and the one-time AI model download are served over the network, and standard anonymous usage stats (such as how many people started a transcription) may be counted — but never the contents of your file, transcript, or edits. We don't make absolute security guarantees beyond how the tool actually works.
Transcribing a YouTube or web video
This tool doesn't take video URLs — there's no way to fetch another site's media from your browser, and we don't run download servers. Instead, use Record tab audio: play the video in another tab, and your browser captures that tab's sound locally (like recording from your own speakers, but digitally clean), then transcribes it on your device as usual. It records in real time — a ten-minute video takes ten minutes of playback — and works with anything that plays in a browser tab. Capture requires desktop Chrome or Edge, and remember to tick "Also share tab audio" in the picker. Please only transcribe content you have the right to use — your own recordings, content licensed to you, or material where transcription for personal use, accessibility, or study is permitted.
Does it work offline?
After the AI model files have finished downloading and your browser has cached them, transcription can run without a network connection. The first run needs to fetch the model, so it isn't offline until that's done.
Tips for the best results
Clear audio transcribes best. Choose Balanced for most recordings; use Fast on phones or older laptops, and Accurate when you have time and want the cleanest text. If your browser supports WebGPU, transcription runs noticeably faster. Very long files take longer and use more memory — for multi-hour recordings, transcribing in sections is smoother.