Genius-Society/soundfonts

Intro

The SF Soft Soundfont Dataset is a comprehensive sample library designed for intelligent music synthesis and digital audio workstation (DAW) development, featuring a diverse collection of high-quality instrument sounds spanning multiple categories including piano, strings, woodwinds, brass, percussion, electronic synthesizers, and ethnic instruments, while supporting both mainstream soundfont formats: SF2, which utilizes uncompressed PCM storage to ensure maximum fidelity and tonal authenticity, and SF3, which employs Ogg Vorbis compression to achieve reduced file sizes and faster loading times; each soundfont is accompanied by structured metadata encompassing instrument classification, pitch range, sample rate, polyphony count, velocity layering, and loop markers, enabling advanced applications such as automatic arrangement, real-time performance, timbre transfer, and cross-platform sequencer development, making it well-suited for mobile music applications, game audio engines, and AI-driven composition systems to facilitate high-fidelity, low-latency intelligent music generation experiences.

Usage

from datasets import load_dataset

ds = load_dataset(
    "Genius-Society/soundfonts",
    name="default",
    split="train",
    cache_dir="./__pycache__",
)
for i in ds:
    print(i)

Maintenance

git clone git@hf.co:datasets/Genius-Society/soundfonts
cd soundfonts

Mirror

https://www.modelscope.cn/datasets/Genius-Society/soundfonts