ReelToText 100% in-browser

About ReelToText

ReelToText is maintained by an independent developer.

What this site does

ReelToText turns a video or audio file into text you can copy or download as subtitles. Seven tools sit on this site, built on the same engine: the TikTok transcript generator, which turns a TikTok video into text; the YouTube video transcript generator; the Instagram transcript generator, which does the same for Instagram Reels and video; the MP3 to text converter, which takes audio with no video attached; the podcast transcription page, which takes an episode file and answers the questions an episode brings — what the intro music does to the first cue, and what happens at a change of speaker; the voice memo to text page, which takes the file a phone wrote and shows the recording level below which a memo stops transcribing; and the subtitle converter, which starts from a recording and shows what the cue timings and line lengths in the subtitle file it produces actually look like. All seven write out plain text with timestamps and can export TXT, SRT or VTT.

None of them is limited to the platform in its name. Any video or audio file the browser can decode will work — a screen recording, a podcast episode, a voice memo, a downloaded clip.

Your file is processed in your browser and is not uploaded

The file you choose is read, decoded and transcribed inside your own browser, and it is never sent to a server. There is no backend on this site that could receive it: the transcription runs locally through WebAssembly, and the only file the page downloads to do that work is the speech-recognition model. Two analytics scripts load alongside it — Cloudflare Web Analytics and Google Analytics — which count visits, carry none of your file, and are named in the privacy policy.

How it is built

The speech recognition engine is Whisper, the open model published by OpenAI. The browser build comes from the open-source Xenova/Transformers.js project, running the model through ONNX Runtime Web compiled to WebAssembly. The three models the tools offer by default are served from this site's own address the first time you transcribe, and your browser caches them, so later runs start faster. The largest option in the model menu is not stored here, and picking it fetches the weights from a third-party model host instead.

What this site is not

ReelToText is not affiliated with TikTok, Instagram, Meta or YouTube, and it is not affiliated with OpenAI, Hugging Face or the Transformers.js project. Platform names are used only to describe what the tools are for.

Cost

All seven are free and need no account. Running the model costs nothing to serve, because the work happens on your device rather than on a server. Advertising may be added later to cover the cost of the domain and hosting; the privacy policy explains what that would involve.

Corrections and questions

If something on this site is wrong, or a tool does not work the way this page says it does, that is worth fixing. The contact page has the address.