About AudioTranscribr

What it is

AudioTranscribr turns audio and video into transcripts, subtitles and a formatted PDF, with a short summary up top. You upload a file or paste a link, and a few minutes later you download the results. The same service is available to developers as a REST API and to AI assistants as an MCP server, so a transcript can be requested from a dashboard, from code, or by asking Claude or ChatGPT.

It does one thing. There is no meeting bot, no video editor and no human transcription marketplace. That focus is why a two-hour recording costs cents here and why a 5 GB video is fine. The first version of AudioTranscribr dates from 2022; the current service is a 2026 rebuild with a different speech engine, a dashboard for people who are not developers, and an API and MCP server for people who are.

Who is behind it

AudioTranscribr is built and run by Ukonnecta LLC, a small independent software company in Marietta, Georgia. There is no sales team and no outsourced support: the people who wrote the service read the support mailbox.

How it works

Speech recognition is done by Deepgram’s Nova models, which we chose after testing on long, multi-speaker public-domain recordings. Our part is everything around it: accepting very large files, extracting audio from video, speaker labels, the five output formats, the dashboard, the API, metering and billing.

Your audio is used only to produce your transcript. Uploads are deleted after 7 days and transcripts after 30, or sooner when you delete them. Nothing is used to train models. The privacy policy says exactly what is stored and for how long.

Proof, not promises

The sample output on the homepage is a real job: a 76-minute Supreme Court oral argument with several speakers, published exactly as the service produced it. The status page reads the same monitoring we do.

Get in touch

Questions, problems or a feature you need: info@audiotranscribr.com. Developers can start at the developer page or the API reference.