Cover of Daydreams; Reverie, sheet music published in New Orleans in 1883
Edward O. Eaton, Daydreams; Reverie, New Orleans, 1883. Representative historical New Orleans material—not a project holding. Library of Congress, Music Division. Public domain. View source →

From fragile object to usable archive

Newspapers, sheet music, letters, photographs, and recordings carry different forms of damage and different kinds of knowledge. The research asks how a source can be captured, recognized, contextualized, and made accessible without separating it from provenance, language, rights, and community authority.

This sheet-music cover is evidence of the problem: image, typography, publication history, place, and musical content must remain connected. A transcription alone is not the archive.

From source to governed access

Source

historical document · image · recording

Capture

digitization · scanning · audio transfer

Recover

image enhancement · layout recognition · restoration

Recognize

OCR · handwriting recognition · speech and audio processing

Contextualize

metadata · provenance · language · dialect · historical usage

Analyze

multimodal search · cultural analysis

Govern & Access

permissions · community input · preservation · reuse

What technology struggles with

faded ink
damaged pages
complex newspaper layouts
handwriting
historical typefaces
multilingual materials
mixed text/image/music documents
degraded audio

Language is part of the archive

Preserving a document is not enough if the language inside it is misread, normalized away, or stripped of local context. Archival Intelligence is studying Louisiana French and other Francophone materials, Kouri-Vini/Louisiana Creole where relevant, historical spelling and orthography, dialect variation, and code-switching.

The project examines the uneven performance of OCR, handwriting recognition (HTR), and speech recognition (ASR) on underrepresented language varieties. Preservation is not merely transcription or translation: language must remain connected to metadata, search, interpretation, and cultural context. This work is developing with Nous Foundation in New Orleans.

Making a document legible is not the same as making history intelligible.

What the machine sees

scan
enhanced image
OCR / HTR / ASR output

What the archive knows

date · creator · place · language · collection · provenance · rights · historical context · relationships

Access is a research question

Permissions, community input, preservation, and responsible reuse are part of the pipeline rather than an afterthought. The people whose histories are contained in an archive should help determine the terms of access and representation. Archival Intelligence is a signatory to the Open Heritage Statement.

Archival Intelligence is an independent research initiative supported by Schmidt Sciences’ Humanities and AI Virtual Institute, with research conducted in collaboration with Kenyon College, Berklee College of Music, Louisiana State University, and other partners.

This approach grows out of broader work at the Human-Centered AI Lab on how AI systems interact with language, culture, judgment, and human values.