Datasets:
Dataset Viewer
The dataset viewer is not available for this dataset.
Unexpected token '<', "<html>
<h"... is not valid JSON
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
txt2vst Dataset
10,000 natural-language-to-VST-spec pairs for music plugin generation.
Dataset Description
Each sample maps a natural language description of a VST instrument to a structured spec.json that defines the complete plugin architecture.
Fields
prompt: Natural language description (e.g., "drum machine with kick snare and acid bass, punchy mastering")completion: Compact JSON spec defining the plugin (voices, FX, theme, mastering chain)
Coverage
| Component | Options |
|---|---|
| Drum voices | kick, snare, hats, tom, perc, clap |
| Pitched voices | bass_acid, lead, pad, pluck, organ, fm_synth, noise, string, brass, sub_bass |
| FX | delay, reverb, chorus, compressor, distortion, phaser, eq, gate |
| Themes | 24 (midnight, void, obsidian, acid, neon, glow, strobe, matrix, ember, solar, copper, candy, frost, chrome, arctic, vapor, industrial, terminal, hologram, white, cream, blood, lavender) |
| Mastering | bypass, transparent, punch, wet, radio, distorted, wide |
Combinatorial Space
- 10M+ sonically unique combinations
- 240M+ visually distinct configurations
- 16 prompt templates with natural variation
Usage
from datasets import load_dataset
ds = load_dataset("fabriziosalmi/txt2vst", split="train")
print(ds[0])
Generation
python3 dataset/generate_dataset.py --count 10000 --output dataset/train.jsonl
License
MIT — same as the txt2vst engine.
Links
- Downloads last month
- 13