Project · 2025

memozero

Own product: local, offline transcription from spoken word to a defensible record. Precision, synced correction, and privacy by design for forensic, legal, and research workflows.

Status
active
Kind
Own
Role
Co-founder and product designer, together with David and Paul, who leads forensics.
Link
www.memozero.io/en

Context

Some recordings must never end up in a cloud: interviews for forensic assessments, witness statements, research data. memozero turns them into text on the user’s own computer. The app runs fully offline on Windows and macOS, recognises who is speaking and keeps a timestamp for every word. No data leaves the machine, so there is nothing to sign with a cloud provider. That matters most for people bound by professional secrecy, who often cannot use cloud services at all.

The website

  • memozero website hero with the transcript editor
    Defensible record
  • Benefits: forensic precision, security, correction and export
    Benefits
  • Local transcription steps from audio file to finished record
    How it works
  • Interactive demo with sample interviews and synced transcript
    Try it

A record that holds up

A transcript that goes into an assessment or a court file has to hold up. memozero plays the audio in sync with the text and highlights each word as it is spoken. It marks passages where the recognition is unsure, and a custom dictionary teaches it the terms of a field. Corrections stay tied to the timeline, and results export to Word, plain text, JSON or Markdown. Every step is designed so the record can be checked later, line by line, against the original audio. The interface is available in English and German. The domain side is led by Paul Jäckel, a forensic psychologist; I design the product together with David.