Greek Council Speech Recognition

Mon, 01 Jan 0001 00:00:00 +0000

Fewer hand corrections for the volunteers who review Greek council transcripts: a Whisper LoRA model trained and published for OpenCouncil during Google Summer of Code 2026. Built the evaluation that decides whether a new model earns its place in production. Deliverable submitted in August 2026; research continues.

The problem Link to heading

Council transcripts need human correction. Model changes must preserve speech and be evaluated consistently before they can justify a change to the transcription workflow.

Teaching a machine to hear a Greek municipal council

Tue, 21 Jul 2026 00:00:00 +0000

This is an English adaptation of a post I originally wrote in Greek for the Schema Labs Substack. The original is here: Μαθαίνοντας μια μηχανή να ακούει ένα ελληνικό δημοτικό συμβούλιο.

OpenCouncil transcribes Greek municipal council meetings with ASR (Automatic Speech Recognition), and then someone fixes the AI’s mistakes by hand. The team spends several hours on each meeting to get those corrections right, so the final transcript matches the audio. That led us to a question: can we fine-tune an open ASR model on OpenCouncil’s own data and cut down the time spent on manual corrections? I’m Angelos, and this is the problem I’m working on for my Google Summer of Code project.

Whisper on Angelos Papamichail

Greek Council Speech Recognition

The problem Link to heading

Teaching a machine to hear a Greek municipal council