VaultCloud AI

Speech to text

Transcribe audio or video into text and subtitles with Whisper.

  • Runs on your device, nothing uploaded
  • Free, no sign-up, no watermark
  • Whisper AI model downloads once

Before you start

These tools are free for everyday use. Please only use files you have the right to use, and don’t use them to mislead, harass or harm anyone, for example by changing documents or IDs, or editing photos of people without their consent.

To keep things that way, images are checked on your device for explicit or violent content, and extracted text is checked for threats. The checks run in this tab, so nothing is uploaded.

Drop an audio or video file

MP3, WAV, M4A, OGG, FLAC, MP4, WebM and more, up to 300 MB.

How to use it

  1. Choose a model. Fast suits clear English speech, Accurate is better for accents and other languages.
  2. Drop an audio or video file. The first time, the model downloads to your browser.
  3. Read the transcript as it appears, then copy it or download it as text or SRT and VTT subtitles.

Good to know

  • Uses OpenAI's open-source Whisper models, running on your device through ONNX Runtime. Accurate downloads a larger model than Fast.
  • Speed depends on your device. With a graphics card (WebGPU) an hour of audio can take a few minutes. Without one it takes longer.
  • Translate to English turns speech in other languages into an English transcript.
  • Indian languages work, with better results from the Accurate model and clear recordings. Check names and numbers carefully.
  • Each part of the transcript is checked on your device for threats before it is shown.
  • Video files are read for their audio track. If a format will not open, convert it to MP3 or WAV first.

Questions

Is my recording uploaded?

No. The audio is decoded and transcribed inside this browser tab. Only the Whisper model is downloaded, once, from Hugging Face.

How long can the recording be?

Files can be up to 300 MB, which is several hours of audio. Long recordings are processed in parts, and you see the text as each part finishes. On phones, keep files under about 30 minutes.

Can I make subtitles?

Yes. Download the transcript as an SRT or VTT file, which YouTube, VLC and most video editors accept.

Can I record people with this?

This tool only transcribes files you already have. Only transcribe recordings you made yourself or have permission to use. See the terms of use.

This tool uses an AI model. Its output can be wrong or incomplete, so check results before relying on them. See the disclaimer.

More tools