AI Subtitle Generator — Whisper Running On Your Own Machine

MediaVault downloads an OpenAI Whisper model into your browser and transcribes your video locally. The audio track never leaves your device, there is no per-minute transcription bill, and the result exports as a standard SRT file you can drop into any editor.

Open the AI Subtitle Generator →

How it differs from cloud captioning

Services like Rev, Otter and Descript upload your recording to their infrastructure, transcribe it there, and retain it under their terms. MediaVault fetches the model weights instead of sending your audio — the direction of transfer is reversed, which is what makes it viable for confidential material.

Who local transcription is for

Anyone whose recordings carry obligations attached to them, plus creators who simply caption enough video that per-minute pricing adds up.

  • Legal teams handling privileged client recordings
  • Clinicians and researchers bound by patient confidentiality
  • Corporate teams under NDA or handling unreleased material
  • Creators captioning weekly video who do not want a metered bill

Accuracy and model sizes

The free tier loads the Tiny English model (roughly 75MB) which handles clear speech well. Pro unlocks the Base model (roughly 140MB), which is noticeably better on accents, background noise and technical vocabulary. The model downloads once and is cached by your browser for later sessions.

How to use it

  1. Load your video. Open the AI Auto-Subtitler and select the video or audio file you want captioned.
  2. Load the Whisper model. Choose a model size and let the weights download into your browser — this happens once.
  3. Export your SRT. Review the transcript and export it as an SRT file ready for any editor or platform.

Frequently asked questions

Is my audio sent to OpenAI or any other server?

No. The Whisper model is downloaded to your browser and inference runs on your device. Your audio is never transmitted.

What subtitle formats can I export?

Transcripts export as SRT, and the subtitle converter turns SRT into VTT, ASS or plain text.

Does it work offline?

Once the model weights are cached by your browser, transcription runs without a connection.

How accurate is it?

Clear, single-speaker audio transcribes very well on the Tiny model. Noisy recordings, strong accents or heavy jargon benefit meaningfully from the larger Base model.

Open the AI Subtitle Generator →

Related tools