Human Vocality Primitives
Controlled airflow textures including exhalations, fricative noise, and shaped airstream bursts.
Structured at the articulation level using documented production workflows and secured under the Proteus Standard™.
Breath and noise airstreams are airflow-driven respiratory behaviors produced without vocal fold engagement, relying instead on controlled inhalation and exhalation, pressure modulation, and open vocal tract configuration to generate sound. In these techniques, acoustic energy is created exclusively through turbulent airflow rather than phonation or articulation, resulting in pure breath-noise output with no pitch, voicing, or linguistic shaping. Expressive control emerges through modulation of airflow intensity, effort, continuity, interruption, and transitions into and out of silence rather than changes in loudness, pitch, or articulatory gesture.
Acoustically, breath and noise airstreams occupy a broadband, noise-dominant timbral space characterized by diffuse spectral energy, intensity-dependent texture, and fine-grained temporal variation. Expressivity is conveyed through airflow pressure, dynamic contouring, interruption patterns, and the contrast between inhalation and exhalation rather than through vocal tract resonance shaping. These characteristics make breath and noise airstreams particularly well suited for modeling non-phonated respiratory sound, effort-aware systems, and airflow-centric acoustic analysis, where precise control of intensity dynamics, gating behavior, and repeatable breath gestures is essential.
Key technical details for this dataset — including file counts, duration, delivery format, and session context.
Planned technical specifications and recording standards for this dataset.
Total Files:
40
Total Files (Preview):
Total Duration (Hours):
Sample Rate (Hz):
96000
Bit Depth (Delivery):
24
Dataset Version:
v1.0
Recording Environment:
Treated Studio
Microphone Configuration:
Rode NT1-A positioned 3-4 inches from mouth
Performer:
Blake Pullen
Recording Dates:
Jan. 30th, 2026
Recording Location:
Las Vegas, NV
Produced using standardized capture, editing, and QC protocols with versioned metadata and Proteus-backed provenance.
An overview of what’s included in this dataset — from articulations and performance styles to session context and recording notes.
This dataset contains a comprehensive collection of recordings capturing non-phonated breath noise and airflow-based respiratory gestures, recorded in isolation to emphasize airflow-driven sound without vocal fold engagement. The material is designed to document the structural organization, capture quality, and gesture taxonomy of breath and noise airstreams as foundational airflow primitives, rather than convey voiced sound, semantic content, or linguistic meaning.
Included recordings emphasize sustained exhaled airflow, dynamic intensity changes, audible inhalation noise, interrupted and gated breath patterns, effort-based pressure noise, and controlled transitions into and out of silence. All material is presented in a neutral, non-performative context to support airflow modeling, embodied system research, and airflow-centric audio analysis workflows. Together, the dataset provides a structured, repeatable corpus suitable for studying and modeling non-phonated respiratory sound across a range of airflow intensity and temporal conditions.
All audio was recorded in a controlled studio environment using standardized capture, editing, and QC protocols applied consistently across the Harmonic Frontier Audio catalog.
Source material was captured at 32-bit float to preserve full dynamic headroom and minimize quantization artifacts during editing and processing. Final dataset files are delivered as 24-bit PCM for consistency and downstream compatibility.
A single performer and respiratory source were used consistently across all sessions to maintain physiological continuity, airflow stability, and repeatable non-phonated breath behavior.
Vocal source details:
Human respiratory airflow — non-phonated breath and noise airstream techniques
Additional processing was limited to trimming, fade handling, and integrity checks. No creative processing, normalization, compression, or dynamic shaping was applied beyond what was necessary for clean, faithful delivery of the recorded material.
A structured breakdown of the expressive building blocks in this dataset — including articulations, dynamics, transitions, and any extended techniques captured during recording.
Unlike clip- or phrase-based datasets, this dataset is structured at the articulation and gesture level. This enables interpretable control, expressive variability, and human-aligned modeling, but significantly increases production complexity and significantly limits who can produce such datasets correctly at scale.
This dataset includes a structured set of non-phonated breath and airflow-based respiratory articulations, recorded in isolation to support airflow-aware modeling and analysis of respiratory sound behavior.
Articulations include:
• Sustained exhaled breath noise at varying airflow intensities
• Dynamic airflow gestures including crescendos, decrescendos, and continuous swells
• Audible inhalation breath noise at controlled intensity levels
• Interrupted and gated airflow events with brief pauses
• Effort-based and burst-style breath noise produced without vocal fold engagement
Articulations are recorded without voicing, linguistic content, accompaniment, or rhythmic framing to preserve clarity, separability, and modeling utility across airflow modeling, embodied systems research, and airflow-centric analysis contexts.
This dataset includes a focused set of gesture-level respiratory behaviors capturing non-phonated breath and noise airstream techniques, with an emphasis on airflow control, pressure modulation, and transitions at the boundary between active airflow and silence.
Gesture types include:
• Micro-variation in airflow intensity and breath stability during sustained exhaled and inhaled airflow
• Dynamic increases and decreases in airflow shaping continuous breath noise gestures
• Subtle changes in respiratory effort affecting noise texture and intensity distribution
• Controlled interruptions and resumptions of airflow, producing gated breath events
• Near-threshold respiratory behaviors where airflow, breath noise, and room tone converge
Gestures are recorded in isolation and without linguistic framing to preserve clarity, repeatability, and modeling utility, supporting detailed analysis of non-phonated respiratory sound and airflow-driven sound generation across airflow modeling and embodied systems research contexts.
This dataset was recorded, documented, and released under The Proteus Standard™, Harmonic Frontier Audio’s framework for rights-cleared, provenance-audited audio data.
The Proteus Standard ensures:
•Performer-owned, contract-clean source material
•Transparent recording methodology and metadata
•Consistent capture, QC, and documentation practices across the catalog
Learn more about The Proteus Standard
Captured with expert musicians and vocalists across global traditions — ensuring each dataset carries authentic nuance, human expression, and rights-managed provenance.
Blake Pullen is a multidisciplinary vocalist, musician, and recording engineer with a background spanning formal vocal performance, traditional acoustic music, and high-fidelity audio production.
With formal training in vocal performance and extensive experience recording both instruments and extended vocal techniques, Blake approaches dataset creation from a physiological, acoustic, and systems-oriented perspective. His work emphasizes articulatory precision, repeatability across sessions, and capture practices designed to support long-term machine learning, speech research, and expressive voice modeling rather than performative presentation.
As the founder of Harmonic Frontier Audio, he performs and records the initial datasets to establish a consistent technical and methodological foundation for the catalog, ensuring that vocal capture techniques, articulation taxonomy, and provenance standards are applied rigorously from the outset.
A three-part listening benchmark: a mixed musical demo built from this dataset, the raw source clip, and an AI model’s attempt to reproduce the same prompt.
A musical demonstration created by replacing a state-of-the-art AI-generated lead instrument with original source recordings from this dataset, then arranged and mastered to preserve musical context. This approach allows direct comparison between current-generation model output and real, rights-cleared acoustic source material.
Directly from the dataset: an isolated, unprocessed example of the source recording.
An unmodified output from a current-gen AI model given the same musical prompt. Included to illustrate where today’s systems still differ from real, recorded sources.
AI model approximations generated using publicly available state-of-the-art music generation systems.
Harmonic Frontier Audio datasets are licensed directly to research teams, startups, and enterprise partners. Access models and terms vary based on use case, scale, and integration needs.
All datasets are delivered with versioned metadata, documented workflows, and Proteus-backed integrity manifests.
We typically respond to inquiries within 1–2 business days.
This dataset is actively being recorded and prepared. You can request early access, previews, or discuss licensing timelines.
We typically respond within 1–2 business days.