Contributing Authors: Moritz Meier, Rathi Adarshi Rammohan, Dennis Küster, Tanja Schultz

EASELAN is a comprehensive human-in-the-loop framework for high-dimensional sensor data annotation aligning multiple modalities including audio, text, video, and biosignals resulting from human behavior and their activities of the eyes, the body, the skin, the muscles, andthe brain. Built upon the well-established ELAN-tool to streamline integration, synchronization, formatting, processing, transcription, annotation, access, sharing, and preservation of signals from audio, text, video, and biosignals. EASELAN is an open-source framework that leverages Git for version control and Continuous Integration/Continuous Deployment pipelines for annotation, validation and automated feedback.

• Supports audio, video, text, and biosignal integration within the ELAN annotation environment.
• Ensures synchronized annotations across different data types.

• Audio & Video Processing: Converts media files for ELAN compatibility using FFMPEG.
• Automatic Transcription: Uses OpenAI’s Whisper for highly accurate speech transcription.
• Biosignal Data Processing: Prepares physiological signals (e.g., ECG, EMG, EEG) for annotation.

• Prepares .eaf files to integrate multiple data modalities.
• Ensures synchronized annotation with ELAN for seamless data alignment.

• Synchronized visualization of high-dimensional sensor data, including EEG, EMG, ECG, IMU, etc., along with audio, video and their corresponding transcriptions

• Uses Git repositories for collaborative annotation tracking.

• Includes continuous verification of annotations:
  • Spell-checking transcripts.
  • Validation against predefined templates.
  • Detection of empty annotation tiers.
• Provides feedback in HTML format, hosted via GitHub Pages.

The src/easelan/ directory contains scripts for pre-processing, annotation, and version control. The CI/CD workflow automates validation and generates annotation reports.