Making history accessible: Doug Boyd’s CURATE research

“Oral history is one of the greatest documents that can be created,” said Doug Boyd, Ph.D., director of the University of Kentucky Libraries Louie B. Nunn Center for Oral History. “It puts individuals on the historical record, and that is really a remarkable thing.”

Boyd has been reimagining how those historical records are accessed. Traditional oral history often sat on library shelves for decades without being heard, but Boyd’s research focuses on the intersection of technology and community to ensure these voices are discoverable and used by the public.

A central component of Boyd’s work to enhance access is indexing — a process that allows researchers to search through hours of audio for specific concepts and stories rather than just keywords. Why this matters: “Somebody could talk about living under segregation for three hours and never once mentioned the word segregation,” Boyd said.

Boyd created OHMS (Oral History Metadata Synchronizer), a web-based, open-source system designed to make oral history searchable and discoverable. Before OHMS, transcripts and audio files often existed in separate silos. Boyd’s system linked them, allowing a user to search for a specific word and jump directly to that moment in the recording. By mapping natural language to concepts and even GPS coordinates, Boyd’s systems provide a level of nuance that traditional transcripts cannot.

The challenge facing the center is the massive scale of its collection, which now includes more than 21,000 interviews. To address this, Boyd utilized his research to move beyond a model where only archivists process collections. This shift was solidified through a grant from the Office of the Vice President for Research’s Celebrating University Research Across the Enterprise (CURATE) program.

The support allowed Boyd to explore a new partnership model: training external communities to index their own stories. In a pilot project involving more than 1,400 interviews with returned Peace Corps Volunteers, community members were trained to use the Nunn Center’s tools to process their own history.

“Engaging with that community has been really transformational,” Boyd said. Because community members act as insiders, they bring terminology and a direct connection to the content that even trained archivists or graduate students might lack. For the Peace Corps project, volunteers have already created 400 indexes, helping to manage a collection that had grown too large for the center to sustain alone.

“In some ways, that has saved the project,” Boyd said.

Boyd is now looking to the future by integrating artificial intelligence through a suite of tools called SpeakEZ, designed to automate the heavy lifting of archival processing. Developed as an in-house, closed system to protect the privacy and sensitive nature of the interviews, SpeakEZ is comprised of five primary modules:

• The Verbatimizer: Utilizing the Whisper model, this tool generates a raw transcript, outputting it in a format that is immediately interoperable with OHMS.
• The Describalizer: This module acts as a digital cataloger, extracting keywords and generating lengthy, two-paragraph archival descriptions. It uses named entity recognition to identify people, organizations and locations, even automatically harvesting GPS coordinates for mentioned places.
• The OHMSifier: This tool creates a draft OHMS index by summarizing the interview at the segment level.
• The Synchronifier: Designed to address a major challenge, this tool adds time codes to old human-generated transcripts that lack time codes and automatically aligns them with the original audio.
• The Riskalizer: Perhaps Boyd’s favorite module, this tool flags potential privacy risks, such as admissions of crime, personal attacks or sensitive medical information, before an interview is ever made public.

While these tools provide efficiency, Boyd emphasizes that they are meant to assist, not replace, the human element. Every AI-generated draft of a transcript and an index is vetted by a human to ensure accuracy and nuance.

Boyd’s CURATE research has reinforced his belief that technology should enhance, not replace, human connection. He remains committed to a model where AI handles the backlog while communities continue the mindful, reflective work of listening to their own stories.

“Anything we can do to enhance engagement with the communities that are represented by the collections that we have — I think that’s just gold,” Boyd said. “I think that’s just the point to which we should be striving.”