Under the hood
How MediaDocks works
The speech tools process your file on the server in a few stages. Most of the image and document tools never leave your browser. This page explains both so you know where your file goes.
Speech to text
How the transcription side works
The speech tools take a file you upload. By default uploads are limited to 100 MB and two hours.
- 1
Normalise the audio
Whatever you upload is decoded and resampled to 16 kHz mono. Speech recognition does not need stereo or high sample rates, so this step shrinks the payload without hurting recognition.
- 2
Split into pieces
Long recordings are cut into sequential segments small enough to send to the speech model. This is why a long lecture spends a while in a processing state: it is handled piece by piece, not in one request.
- 3
Transcribe and stitch
Each segment is sent to Google's Gemini API for recognition, then the returned text is reassembled in order. The spoken language is detected automatically instead of being something you have to declare.
- 4
Time-align for subtitles
When you ask for subtitles, the recognised text is paired with timings and written as cues, then offered as SRT for editors and social platforms or WebVTT for the browser's native track element.
Accuracy is better on clear single-speaker audio and drops with background music, crosstalk and heavy echo. Treat a transcript as a first draft. The guide to how speech recognition works explains where the remaining errors come from.
On your device
Tools that run in your browser
These tools process your files with code that runs in your browser tab. The files are not uploaded to a MediaDocks server.
Image Converter
Converts between common image formats on your device.
Image Resizer
Resizes by pixels or percentage on your device.
Background Remover
Runs an image model in your browser after fetching it once.
PDF to Image
Renders PDF pages to images on your device.
Image to PDF
Combines images into a PDF on your device.
SVG to PNG
Rasterizes vector files on your device.
Base64 Image
Encodes and decodes image data on your device.
QR Code Generator
Draws QR codes on your device.
Two tools do use the server: Video to Audio uploads your video so the audio can be extracted, and the URL Shortener stores the links you create. Details are in the privacy policy.
Pick a tool and start with your own file.
Transcribe a recording, add subtitles, convert an image or document, or read a guide first. There is nothing to install and no sign-up step.