Apply for class 03 now!

Transcribee-Desktop: Transcribe audio and video recordings locally

About the project

Team Members

Philipp Mandler, Jaro Habiger, Robin Heinemann

Funding Period

In funding since 01/06/2026

What is the project about?

Creating audio transcripts requires a tremendous amount of effort, especially for longer audio or video recordings. That is why AI tools are being used more and more frequently, although their use often comes with data privacy concerns. Furthermore, fully machine-generated transcripts are not good enough for many applications and lead, for example, to so-called “craptions”, that is, automatically generated subtitles of poor quality, so that accessibility issues are masked rather than resolved. With Transcribee Desktop, we are creating the local counterpart to Transcribee, a web platform for audio transcriptions. Transcribee Desktop can be used by individuals without requiring additional infrastructure or specialized knowledge, without producing problems for data protection.

Which audience does the project address?

Transcribee is already used by a variety of users, primarily in research and civil society. Transcribee Desktop is now intended to reach out to research projects, students, journalists, and other target groups who currently cannot, are not allowed to, or do not wish to use Transcribee for data protection or organizational reasons. In addition, Transcribee Desktop offers a low-threshold, privacy-savy alternative to common cloud services used for transcription for everyday use.

What is to be achieved?

Transcribee Desktop is intended to become a cross-platform app capable of processing local files. At the same time, Transcribee’s existing codebase is being revised, and program libraries are being developed so that they can be used by Transcribee, Transcribee Desktop, and third parties. In addition, recent research findings on optimizing automatically generated transcripts are expected to lead to improved quality.