New Dataset Release from Harmonic Frontier Audio (HFA)
Subharmonic Phonation / Vocal Fry (Preview)
View on Hugging Face →
We’re excited to share the latest release in the Harmonic Frontier Audio – Extended Vocal Techniques Series: a high-fidelity dataset exploring subharmonic phonation, one of the world’s rarest and most fascinating vocal phenomena.
What it is
Subharmonic phonation (commonly referred to as vocal fry when produced at low frequencies) is a complex vocal technique in which the vocal folds oscillate at frequencies that are integer divisions of the fundamental pitch, creating a characteristic growling, overtone-rich sound. This technique produces a lower pitch, most commonly an octave below the sung pitch, by allowing the lower frequency to occur alongside the primary one.
This form of voicing is present in numerous world singing traditions — including Tuvan and Mongolian throat singing, Tibetan Buddhist chant, and various forms of overtone or harmonic singing.
It is also occasionally employed in Western vocal performance, experimental music, and cinematic sound design for its unique texture and resonance.
This dataset presents a neutral, non-traditional representation of the technique — recorded as a general-purpose study of subharmonic phonation rather than as an emulation of any specific cultural or regional style.
The intention is to provide AI researchers and sound designers with clean, controlled examples of this complex human sound source for analysis, synthesis, and expressive modeling.
Technical Details
-
Recorded and performed by Blake Pullen
-
96 kHz / 24-bit mono WAVs
-
12 files (sustains, glissandi, and dynamics)
-
Captured via Rode NT1A → Zoom F8n Pro
-
Includes structured metadata, rights declaration, and README documentation
-
Licensed under CC BY-NC 4.0 (non-commercial use)
Listen
Hear the dataset in a short mixed demo below:
(The instrumental backing was generated using Suno’s Pro plan under a commercial license; all vocal content and featured dataset sounds were performed and produced by Blake Pullen for Harmonic Frontier Audio.)
About Harmonic Frontier Audio
HFA creates rights-cleared, provenance-verified datasets of world instruments and extended vocal techniques for ethical AI training and creative research.
Each dataset is published with transparent metadata, technical provenance (The Proteus Standard™), and clear licensing.
We’d love your feedback.
-
Does this kind of dataset help your work in voice synthesis or expressive-audio modeling?
-
Which other extended vocal techniques would you like to see next?
Feel free to comment below, share examples, or reach out via info@harmonicfrontieraudio.com