Gratz College Library AI Projects

AI-Assisted Transcription for Holocaust Oral Histories
Gratz College logo

Gratz College Library is the custodian of a unique collection of oral history interviews with Holocaust survivors, recorded during the 1983 American Gathering of Jewish Holocaust Survivors. To support the library's digital initiatives, I developed an AI-powered transcription workflow that automated the transcription of hundreds of unprocessed interviews.

This project involved:

AI Workflow Design

Built a workflow using WhisperX, combining:
- OpenAI Whisper for high-accuracy speech-to-text conversion.
- Pyannote-audio for speaker diarization to distinguish speakers in dialogues.

Automation

Developed scripts in Python to batch process audio files, generating preliminary transcripts in Microsoft Word for archivist review.
Integrated error-checking mechanisms to ensure text alignment with audio timestamps.

Improved Efficiency

Reduced transcription time by over 80% compared to manual workflows.
Enabled archivists to focus on contextual annotations and metadata enrichment.

This project demonstrates my ability to implement AI tools to modernize traditional archival processes, making collections more accessible and usable for researchers.