VocalPrint — The Living Sound Museum

Exhibit 01 · The Crisis

Three Crises.
One Blind Spot.

Every two weeks, a language vanishes. Over 40% of the world's 7,000 languages are endangered, yet fewer than 15% have been digitized. This crisis runs deeper than preservation statistics. It is a crisis of belonging, of infrastructure, and of who AI was built to hear.

The AI Listening Gap

AI Cannot Hear You.

2× the word error rate for Black speakers vs. white speakers

Major speech-to-text systems from Amazon, Apple, Google, IBM, and Microsoft all show significant racial and dialectal disparities. AI now screens job applicants, transcribes medical conversations, and grades students. It mishears half the population while deciding their future.

The Diasporic Wound

A Grief Without a Name.

"Cultural bereavement: the experience of the uprooted person resulting from loss of social structures, cultural values and self-identity."

World Psychiatry, 2005

2nd and 3rd-generation diaspora individuals often cannot speak their heritage language fluently but keep the longing for it. The urgency is temporal: elderly speakers are aging out now. Nobody pays for "preservation." People pay for belonging, recognition, and continuity.

The Corpus Vacuum

Infrastructure Nobody Built.

$71.77B global language services market, mostly textual

Hundreds of endangered languages exist only in oral form. Ainu in Japan, Aromanian across the Balkans, Quechua in Peru, Arrernte in Australia. When a language has no written corpus, the entire localization supply chain breaks down. Whoever builds the verified oral corpus for low-resource languages becomes the infrastructure layer for everyone downstream: AI companies, edtech, fintech, healthcare.

Exhibit 02 · The Vision

Identity Infrastructure,
Not an Archive.

VocalPrint is not a preservation project, i.e. a shelf, a record, or a photograph of something dying. We are a living museum building on a foundation of community trust. While a competitor can copy technology, they cannot replicate a model rooted in genuine sovereignty and consent. Our commercial viability grows from the very communities that build the corpus.

Phase 1 · Starting Point

The Museum

A consumer platform for the diaspora to preserve the sound of home. This is our first product and the emotional entry point for the community. It is the community-building mechanism that powers everything follows.

Closest Analog

Ancestry DNA for the voice

Revenue

Consumer subscription · Heritage Kit · Diaspora org partnerships

Phase 2 · Commercial Layer

The Corpus

A verified, community-consented, linguistically-tagged voice corpus with full provenance documentation. We provide the "Corrective Data" for AI companies that currently fail accented speakers. Ethical architecture is our competitive moat.

Buyers

ASR companies · HR tech · Healthcare ambient AI · Consumer tech

Revenue

Data licensing · API access · Enterprise contracts · Audit services

Phase 3 · Emergent

The Infrastructure

The long-term infrastructure play for 300+ languages that big tech will never touch. From Tulu in southern India to Sorbian in Germany to Mapuche in Chile, we become the global gateway for any company entering low-resource language markets.

Sequencing

Community trust → Corpus → Commercial layer → Infrastructure moat

Exhibit 03 · The Method

Governance Is
the Product.

We do not build the app first. We build the trust. The consent framework and data model must exist before the first recording is made.

Voice data is biometric data in several jurisdictions. Oral traditions are communally owned, not individually owned. The IP architecture has three distinct layers: contributor owns their voice, platform may own the recorded artifact, community may hold claims on the cultural content expressed. Getting this wrong at the start is irreversible. Getting it right is a moat.

Before the First Recording

1

Contributor Agreement

Plain language, not just ToS. What is collected, how it is stored, who can access it, what commercial uses are permitted or prohibited, the compensation structure, and the withdrawal mechanism.
2

Community Consent Framework

How community-level consent is obtained, who has authority to grant it, what happens when individual and community consent conflict, and how communities can renegotiate terms over time.
3

Data Schema Specification

Every metadata field attached to every recording: language, dialect, region, speaker age, consent permissions, community ratification status, permitted use categories. Designed by a computational linguist, not a product manager.
4

Entity Structure Document

Certificate of incorporation as a Public Benefit Corporation, codifying the mission as a permanent requirement, protecting from future market or investor pressures.
5

Ethics Advisory Council Terms of Reference

Composition and authority of a functional oversight board with veto power, established before data collection to ensure independent governnce.

The Flywheel

Exhibit 04 · The Founder

Shalom Donga

As Chinua Achebe famously said, “Until the lions have their own historians, the history of the hunt will always glorify the hunter.”

I am part of a community that exists across borders and outside the systems most people rely on. We are the Ndebele people, representing roughly 0.05% of the global population.

I study computer science at Grambling State University. Alongside that, I have built production-level Flutter applications and worked on multilingual AI evaluation. That work has made one thing clear to me: what gets modeled is a choice, and what is left out follows a pattern.

I choose to work on problems that extend beyond my immediate sphere, building technology that reaches people it was not originally designed for and making it as inclusive as it can be.

In the words of Toni Morrison, “If you have some power, then your job is to empower somebody else.”

VocalPrint is built with that principle in mind, with governance, consent, and structure treated as foundational decisions rather than afterthoughts.

VocalPrint is in its founding phase. If you are a researcher, builder, or organization working in language, AI, or community infrastructure, or you are just interested, scroll down to the Collaborate section.

Because, as Nelson Mandela reminded us, “If you talk to a man in a language he understands, that goes to his head. If you talk to him in his language, that goes to his heart.”

Exhibit 05 · Collaborate

This Isn't Built Alone.

There is no finished product to join, only the opportunity to shape what gets built. We are designing the governance, consent frameworks, and data architecture that will define VocalPrint from the ground up. The people who engage now will not just inherit the system, but will determine how it works. Find your entry point below.

Legal Counsel

We are looking for legal architects who can help design the foundation, not just advise on it. Voice data is treated as biometric data in multiple jurisdictions, and oral traditions often exist as communal property. This requires expertise across IP law, data protection, and indigenous cultural heritage, alongside experience with data licensing, AI regulation, and hybrid nonprofit-commercial structures. The goal is to build consent frameworks and legal infrastructure that hold up across jurisdictions from the start.

Write to us → Legal

Academic Collaborators

The research questions are inseparable from the product itself. We are working with computational linguists, ethnomusicologists, oral tradition scholars, and AI or ML engineers with ASR specialization to define the data architecture before any recording begins. This includes questions of what makes a corpus commercially viable, how oral traditions function as shared property, and what level of quality and structure is required for AI systems to use it reliably.

Write to us → Research

Community Organizations

We are building with communities as participants in governance, not as end users. This includes diaspora organizations, indigenous language boards, cultural centers, and heritage language schools. If you represent a language community, your role is not to validate decisions after the fact but to help shape how consent, ownership, and participation are defined from the outset.

Write to us → Community

Investors & Co-Builders

Mission-aligned only.

We are looking for mission-aligned partners at the pre-seed stage who understand where the defensibility comes from. The system is not differentiated by the software, but by the structure that governs how data is contributed, controlled, and used. Communities do not contribute voice data without clear terms, and once those terms are formalized through consent frameworks and data agreements, access becomes limited in a way that cannot be easily reproduced. A competitor can build similar technology, but cannot access the same data without rebuilding those relationships and governance structures over time.

Write to us → Investment