elefcode

Speech to Text

Transcribe audio and video into text and subtitles with Whisper AI, running on your own device. Your recording is never uploaded.

Drop an audio or video file here

MP3, WAV, M4A, OGG, FLAC, WebM, MP4 · meetings, lectures, interviews, voice notes

Only transcribe recordings you have the right to use, and ask before recording other people.

Model

The model downloads once from Hugging Face and is kept in your browser; your audio stays on this device. Choosing the language, when you know it, is faster and more accurate.

Advertisement

Guide

About the Speech to Text

This tool turns speech into text with Whisper, the speech-recognition model released by OpenAI. Instead of sending your recording to a server, it downloads the model once and runs it in your browser, so meetings, interviews, lectures and voice notes stay private. The result comes with timings, ready to correct and save as a transcript or as subtitles for your video.

Which model should I choose?

  • Tiny (45 MB) — quickest; good for clear speech in English and other widely spoken languages.
  • Base (82 MB) — the best balance of speed and accuracy for most recordings.
  • Small (254 MB) — noticeably better with accents, noise and less common languages, but several times slower.

Getting the most accurate transcript

Pick the spoken language instead of relying on detection, use the clearest recording you have, and choose a bigger model for difficult audio. Whisper adds punctuation and capital letters on its own. It does not tell speakers apart, so for interviews you may want to add names while you edit the lines.

How to use it

  1. 1Choose an audio or video file, record from your microphone, or try the sample.
  2. 2Pick a model and, if you know it, the spoken language.
  3. 3Click Transcribe — the model downloads the first time, then the lines appear as they are done.
  4. 4Correct any words, then copy the text or download TXT, SRT or VTT.

Frequently asked questions

Is my audio uploaded?

No. Only the speech model is downloaded, from Hugging Face; your recording is processed by your own browser and never leaves your device.

How long does it take?

It depends on your computer and the model. On a typical laptop, Tiny and Base transcribe faster than the recording plays; Small can take about as long as the recording or longer.

Does it support Arabic?

Yes. Whisper understands Arabic and more than 90 other languages. For Arabic, the Small model gives clearly better results than Tiny.

Can I make subtitles for YouTube?

Yes. Download the SRT file and upload it in YouTube Studio under Subtitles, or open it alongside your video in VLC and most editors.

Is there a length limit?

No fixed limit. Long recordings are processed 30 seconds at a time, so an hour-long file works if your device has enough memory — keep the tab open until it finishes.