Partners¶
CLAMS is developed by the Brandeis Lab for Linguistics and Computation (LLC) in close partnership with the GBH Archives, home of the American Archive of Public Broadcasting. The project is funded primarily by the Mellon Foundation, with additional support from the Institute of Museum and Library Services.
GBH Archives¶
The GBH Archives is the archives unit of GBH, the Boston public media producer. It makes GBH’s historic materials accessible to producers, scholars, researchers, educators, students, lifelong learners, and the public. It publishes selected holdings on Open Vault and, together with the Library of Congress, stewards the American Archive of Public Broadcasting (AAPB), which preserves and provides access to historically significant public radio and television programs from stations and producers across the United States.
The GBH Archives has been the CLAMS project’s primary partner since the project began in 2019. GBH archivists define the cataloging problems that CLAMS apps address, contribute training and evaluation data drawn from AAPB holdings, and put the resulting tools to work in their cataloging workflows. Their feedback drives what gets built and how it is evaluated.
The GBH Archives is located in Brighton, Massachusetts, and hosts researchers by appointment. To reach the archives team, email aapb_notifications@wgbh.org, use the GBH contact form, or follow @GBHArchives. For questions about the CLAMS collaboration itself, email admin@clams.ai.
The GBH team writes regularly about this work on the AAPB blog. A few representative posts, from building training data for CLAMS apps to using CLAMS output in day-to-day cataloging:
Beyond the Screen: Machine Learning and Metadata Creation at GBH Archives (December 2024): building training data to help CLAMS turn on-screen text into structured metadata.
Be Not Afr(AI)d: A GBH Intern’s Perspective on the Future of AI in Archives (April 2025): curating data for the scenes-with-text detection app and reporting findings back to the Brandeis team.
Outside the Black Box: AI Shedding Light for AV Cataloging (April 2026): how GBH turns MMIF output into visual indexes that catalogers use during their workflow.
Contributor Cataloging in the American Archive of Public Broadcasting (August 2026): documenting program contributors at scale with CLAMS apps, and the privacy questions that come with it.
Mellon Foundation¶
The Mellon Foundation has supported the AAPB through a series of grants to GBH. CLAMS has been part of that support since 2019, with Brandeis taking part as the developer of the platform.
2019: a two-year, $750,000 grant, From Repository to Resource, which paired the AAPB with the LLC to use machine learning to capture richer metadata from radio and television programs. The CLAMS platform was born under this grant.
2022: a four-year, $16 million grant to preserve endangered public media holdings at scale. Brandeis received an $825,000 subgrant to apply CLAMS to a substantially larger and more diverse collection and to develop workflows that let archivists adapt current AI tools to new material.
2025: a grant for low-resource spoken language identification in the AAPB, with Brandeis again as subrecipient. The project targets Spanish, American Samoan, and Central Alaskan Yup’ik in AAPB holdings, so that multilingual items become discoverable by the communities and researchers who need them. Early results appeared at IASA 2025 and at AmericasNLP 2026.
Institute of Museum and Library Services¶
In 2022 the Institute of Museum and Library Services (IMLS) awarded GBH a National Leadership Grant for Libraries to improve speech-to-text transcripts in digital collections. The project upgraded the AAPB’s Kaldi-based transcription toolkit with modern neural language models and streamlined FIX IT+, the AAPB’s crowdsourcing platform for correcting transcripts. Brandeis took part as a subrecipient, leading the work on the transcription toolkit and presenting the results to the library and archives community.