Fan Composer: NPU-Assisted Fan-Aware Soundscapes with Always-On Automatic Volume

Fan Composer is a local, Rust-native Windows research application that analyzes laptop fan noise and adds a quiet adaptive soundscape. The project began because an Intel Core Ultra 7 laptop's NPU was rarely used in everyday work, motivating a search for a useful local application for that otherwise idle hardware. The application implements Harmonic, Nature, and Generative synthesis, always-on automatic volume bounded by a user-selected maximum, a conservative stereo playback-reference guard, and a 1 Hz OpenVINO music policy with NPU-first execution and CPU fallback. FFT, automatic volume, limiting, and audio synthesis remain CPU DSP. The small neural network imitates locally generated rules; it is not trained on human comfort ratings. The accompanying engineering preprint reports 26 passing automated tests, closed-loop actual-FFT/synthesis simulations of fan-level steps in all three modes, real native NPU inference, CPU fallback injection, and a 45-second silent physical duplex run with 45 NPU policy decisions. The silent run recorded zero input/reference queue drops, zero stream faults and zero callback over-budget events, with one startup input Xrun. These observations do not establish audible comfort, energy savings, general source separation, physical acoustic rejection, or long-session stability. The preprint includes the application screenshot, architecture, automatic-volume method, model provenance, actual results, build/use guidance and proposed perceptual/energy experiments. Source code, model and reproducibility materials are available in the linked GitHub repository. Describes software version 0.3.0. Preprint; not peer reviewed.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-10-20
DOI
https://doi.org/10.5281/zenodo.23478731
Primary Topic
Music Technology and Sound Studies
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
preprint

Fan Composer: NPU-Assisted Fan-Aware Soundscapes with Always-On Automatic Volume

Kyu Lee
Zenodo (CERN European Organization for Nuclear Research)
Music Technology and Sound Studies
preprint

Fan Composer: NPU-Assisted Fan-Aware Soundscapes with Always-On Automatic Volume

Kyu Lee
preprint en

Abstract

Fan Composer is a local, Rust-native Windows research application that analyzes laptop fan noise and adds a quiet adaptive soundscape. The project began because an Intel Core Ultra 7 laptop's NPU was rarely used in everyday work, motivating a search for a useful local application for that otherwise idle hardware. The application implements Harmonic, Nature, and Generative synthesis, always-on automatic volume bounded by a user-selected maximum, a conservative stereo playback-reference guard, and a 1 Hz OpenVINO music policy with NPU-first execution and CPU fallback. FFT, automatic volume, limiting, and audio synthesis remain CPU DSP. The small neural network imitates locally generated rules; it is not trained on human comfort ratings. The accompanying engineering preprint reports 26 passing automated tests, closed-loop actual-FFT/synthesis simulations of fan-level steps in all three modes, real native NPU inference, CPU fallback injection, and a 45-second silent physical duplex run with 45 NPU policy decisions. The silent run recorded zero input/reference queue drops, zero stream faults and zero callback over-budget events, with one startup input Xrun. These observations do not establish audible comfort, energy savings, general source separation, physical acoustic rejection, or long-session stability. The preprint includes the application screenshot, architecture, automatic-volume method, model provenance, actual results, build/use guidance and proposed perceptual/energy experiments. Source code, model and reproducibility materials are available in the linked GitHub repository. Describes software version 0.3.0. Preprint; not peer reviewed.

Zenodo (CERN European Organization for Nuclear Research)
Music Technology and Sound Studies
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.