The foundational acoustic dataset platform — voice, instrument, breath, and human gesture.

Rights-cleared, provenance-audited datasets spanning world instruments, extended vocal techniques, emotional and physiological vocality, and the acoustic primitives of human gesture — engineered for music generation, expressive TTS, multimodal AI, robotics, and synthetic humans.

Structured at the articulation level, produced through documented workflows, and secured by the Proteus Standard™—so as scrutiny around training data provenance and auditability increases, teams can clearly explain not just what data they use, but where it came from and how it can be defended.

Browse the Catalog Enterprise Licensing Inquiries

New here? Start with this.

If this is your first visit, these four pages provide the fastest orientation to what Harmonic Frontier Audio is, how teams use the datasets, and how licensing works.

1. About HFA — What this platform is (and what it isn’t)
‍
2. How to use these datasets — Workflow fit, formats, and integration

3. Data use cases — what teams actually build with HFA data
‍
4. FAQ — Licensing, compliance, and delivery basics

Most teams begin with a short conversation to confirm fit.

Elements of Sound

Unlike clip- or phrase-based datasets, HFA structures sound at the articulation and gesture level—enabling controllable, interpretable model behavior.

AIR

Wind-borne instruments, flutes, pipes, whistles, and airflow primitives shaped by breath and pressure.

STRING

Fiddles and bowed-string traditions that carry melodic contour, ornamentation, and expressive resonance.

SKIN

World percussion, hand and frame drums, struck surfaces, and body-driven gesture and foley primitives.

METAL

Bells, plates, metallic percussion, and sustained harmonic resonances for tone, shimmer, and impact.

VOICE

Extended techniques, overtone singing, subharmonic phonation, emotional vocalizations, and core human vocality primitives.

Catalog Series

Structured dataset families spanning instruments, vocality, breath, and human gesture — each engineered for clarity, expressive range, and provenance.

Celtic Constellation

Traditional and modern Celtic instruments and voices—pipes, whistles, fiddles, and more—captured for melodic, modal, and drone-based generation.

World Percussion Nexus

Frame drums, hand percussion, and rhythmic instruments from around the world—focused on gesture-rich, playable patterns.

Extended Vocal Techniques Spectrum (Music)

Overtone singing, multiphonics, throat voice, and experimental vocal colors presented as phrase-level, musical gestures for expressive voice modeling.

World Resonance Gallery

Plucked, bowed, struck, and wind-resonant instruments—kalimba, wooden flutes, ocarina-type voices, and other acoustic bodies—engineered to teach models resonance shaping, harmonic decay, and spectral color transitions.

Novelty Gems Cabinet

Oddities, curiosities, and rare sound-makers—boutique textures for giving models a unique sonic fingerprint.

Human Vocality Primitives

Foundational non-verbal phonation datasets—breath noise, glottal gestures, and expressive primitives engineered for speech, robotics, TTS, and multimodal AI.

Human Gesture & Body Primitives

Body-driven sounds—clothing movement, footsteps, impacts, and subtle human foley—captured as clean primitives for multimodal, robotics, and video-to-audio models.

Our Principles of Resonance

Five pillars guide everything we create—from how we record to how we license.
Each one reflects a balance between ethics, artistry, and intelligence.

Provenance with Purpose

Every sound we capture is traceable to its source—performer, session, and signature—across instruments, vocality, breath, and human gesture. The Proteus Standard™ ensures every dataset carries verified provenance, transparent lineage, and ethical clarity.

Artistry in Architecture

Data can be beautiful. From waveform to metadata, we design every dataset with the same care an artist gives to composition—balanced, structured, and human.

Cultural Stewardship

Every culture has its sound. We record and preserve rare instruments, vocal techniques, and embodied sonic traditions—ensuring their global stories are heard and remembered in the age of AI.

Human-Aligned AI

Through our Orpheus Suite, we teach AI not only to hear but to listen—across instruments, extended vocal techniques, physiological breath, emotional vocalizations, and human gesture. Our instruction-tuned datasets align machines with the intent, nuance, and non-verbal expressivity of human sound for music, speech, robotics, multimodal learning, and next-generation voice systems.

Simplicity in Discovery

Our catalog is structured for clarity—organized by Series and Sound Families, with expressive primitives and descriptive Elements making the world of sound intuitive to explore.

Producing articulation-level datasets requires the convergence of performance expertise, controlled acoustic capture, and ML-aligned metadata architecture. This combination significantly limits who can produce such datasets correctly, and is the primary reason comparable catalogs are rare.

Featured Datasets

A growing catalogue of rights-cleared acoustic datasets—spanning instruments, vocality, breath, and human gesture—optimized for modern generative and multimodal AI.

Highland Bagpipes

Celtic Constellation · Air

A focused preview of Highland pipe gestures—sustains, embellishments, and calibrated dynamic sweeps for melodic and drone-based generation.