Tool
TransVisEd
A transcript visualization, evaluation and editing workbench for comparing AI-generated transcripts and supporting human validation.
Open demonstrator →Research infrastructure for spoken historical sources
Building the future of AI-enabled oral history research
MDOH is an open research initiative developing methods, tools and interoperable datasets for the analysis, preservation and interpretation of oral history interviews.
About
Oral history beyond the transcript
Oral history interviews contain more than words. Pauses, laughter, overlaps, speaking tempo and other interactional features are essential parts of the historical record.
MDOH brings together artificial intelligence, digital humanities and research data management to support transparent, reusable and researcher-centred workflows for working with spoken historical sources.
Tools and resources
From prototype to research infrastructure
MDOH is designed as a growing ecosystem of tools, datasets and documentation.
Datasets
Benchmark resources
Curated training and evaluation data for transcription, alignment and paralinguistic event detection.
In developmentDocumentation
Methods and workflows
Practical guidance for reproducible AI-assisted processing of oral history interviews.
Coming soonResearch
Current themes
The initiative focuses on methods that preserve the complexity and evidential value of spoken historical sources.
Automatic speech recognition
Evaluation and adaptation of ASR systems for spontaneous historical interviews.
Multimodal oral history
Representation of speech, timing, interaction and paralinguistic information.
Research data infrastructure
Interoperable formats, transparent workflows and sustainable access.
Collaboration
Interested in MDOH?
We welcome collaboration with oral historians, archives, digital humanities researchers and developers.