🛠️ ALPHA SOFTWARE UNDER CONSTRUCTION — Use at your own risk. Backwards compatibility not guaranteed.

╔══════════════════════════════════════════════════════════════════════════════╗
║                                                                              ║
║                     ██╗   ██╗███╗   ██╗██████╗ ███████╗██████╗              ║
║                     ██║   ██║████╗  ██║██╔══██╗██╔════╝██╔══██╗             ║
║                     ██║   ██║██╔██╗ ██║██║  ██║█████╗  ██████╔╝             ║
║                     ██║   ██║██║╚██╗██║██║  ██║██╔══╝  ██╔══██╗             ║
║                     ╚██████╔╝██║ ╚████║██████╔╝███████╗██║  ██║             ║
║                      ╚═════╝ ╚═╝  ╚═══╝╚═════╝ ╚══════╝╚═╝  ╚═╝             ║
║                                                                              ║
║                     ██████╗ ██████╗ ███╗   ██╗███████╗████████╗██████╗      ║
║                    ██╔════╝██╔═══██╗████╗  ██║██╔════╝╚══██╔══╝██╔══██╗     ║
║                    ██║     ██║   ██║██╔██╗ ██║███████╗   ██║   ██████╔╝     ║
║                    ██║     ██║   ██║██║╚██╗██║╚════██║   ██║   ██╔══██╗     ║
║                    ╚██████╗╚██████╔╝██║ ╚████║███████║   ██║   ██║  ██║     ║
║                     ╚═════╝ ╚═════╝ ╚═╝  ╚═══╝╚══════╝   ╚═╝   ╚═╝  ╚═╝     ║
║                                                                              ║
║                     █████╗ ██╗     ██████╗ ██╗  ██╗ █████╗                  ║
║                    ██╔══██╗██║     ██╔══██╗██║  ██║██╔══██╗                 ║
║                    ███████║██║     ██████╔╝███████║███████║                 ║
║                    ██╔══██║██║     ██╔═══╝ ██╔══██║██╔══██║                 ║
║                    ██║  ██║███████╗██║     ██║  ██║██║  ██║                 ║
║                    ╚═╝  ╚═╝╚══════╝╚═╝     ╚═╝  ╚═╝╚═╝  ╚═╝                 ║
║                                                                              ║
║                    ╔══════════════════════════════════════╗                  ║
║                    ║     ⚠️  ALPHA SOFTWARE  ⚠️           ║                  ║
║                    ║  EXPECT BREAKING CHANGES & BUGS     ║                  ║
║                    ║  NOT READY FOR PRODUCTION USE       ║                  ║
║                    ╚══════════════════════════════════════╝                  ║
║                                                                              ║
╚══════════════════════════════════════════════════════════════════════════════╝

Dance Dance Convolution

Modernized pipeline for DDR chart generation and analysis.

Overview

This repository contains a modernized DDC workflow for:

• official DDR corpus download/extraction
• dataset preparation and filtering
• audio feature extraction
• DDC onset + placement-model training
• difficulty-evaluator training
• inference / AutoChart generation
• ArrowVortex-oriented integration work

Recent work in this repository also added:

• .ssc extraction support in addition to .sm
• corpus auditing for note-symbol coverage
• note-object semantic auditing
• refreshed PyTorch-oriented training/inference wiring for the active environment

Installation

1. Clone with submodules:

   git clone --recursive <repo_url>
   cd ddc

2. Install dependencies as appropriate for your environment.
3. Install the package if desired:

   pip install .

Usage

AutoChart CLI

Generate stepcharts for a song or directory of songs:

python autochart.py path/to/song.mp3 --models_dir path/to/models --ffr_dir path/to/ffr_models

For the refreshed local .ssc-inclusive run, the current direct-use paths are:

python autochart.py path/to/song.mp3 --models_dir data/ssc_refresh_work/models --ffr_dir data/ssc_refresh_work/ffr_models

Or for a directory:

python autochart.py path/to/songs/ --models_dir path/to/models

Options include:

• --out_dir: output directory
• --ffr_dir: path to trained difficulty models
• --google_key: optional Google API key for image lookup
• --cx: optional Google Custom Search ID

Library Usage

from infer.autochart_lib import AutoChart

ac = AutoChart(models_dir='models_dir', ffr_model_dir='ffr_models')
ac.process_song('mysong.mp3', 'output_dir')

Training

To run the end-to-end training pipeline:

python scripts/train_all.py /path/to/sm_packs /path/to/work_dir

Current Documentation

Key project analysis/report documents:

• docs/TRAINING_ANALYSIS_2026-04-04.md
• docs/CORPUS_AUDIT_2026-04-04.md
• docs/SSC_EXPANSION_ANALYSIS_2026-04-04.md
• docs/NOTE_OBJECT_SEMANTICS_2026-04-04.md
• docs/TRAINING_REFRESH_COMPLETION_2026-04-04.md
• docs/REFRESH_DEPLOYMENT_AND_ARROWVORTEX_VERIFICATION_2026-04-04.md

Features

• modernized Python 3 workflow
• official DDR pack download support
• .sm and .ssc extraction support
• DDC onset detection and step generation pipeline
• floating-point difficulty regression support
• ArrowVortex-oriented server/inference path

Credits

• Original DDC: Chris Donahue et al.
• Modernization / integration work in this repo: subsequent contributors
• DDC onset port: Robert Pelloni
• FFR difficulty model integration: Robert Pelloni