severyn-k/isolated-guitar-chords

Isolated Guitar Chords Dataset

Overview

This dataset contains isolated guitar chord recordings designed for audio classification tasks such as chord recognition and real-time music analysis.

The data was intentionally recorded in realistic acoustic conditions, including minor background sounds, to improve robustness during inference in real-world environments.

---

Recording Setup

• Instrument: 🎸 Fender FA-15 3/4 Acoustic Guitar
• Recording method: Manual recording
• Environment: Quiet room with an open window
• Background noise: Minor ambient sounds intentionally allowed
• Average duration per file: ~5.5 seconds

This setup produces high-quality yet realistic recordings, reflecting conditions similar to those expected during deployment (e.g. home practice, casual recording environments).

---

Dataset Structure

data/
├── train/
│ ├── A/
│ ├── Am/
│ ├── A#/
│ ├── ...
│ └── Noise/
├── test/
│ ├── A/
│ ├── Am/
│ ├── A#/
│ ├── ...
│ └── Noise/

• Each chord class is stored in its own directory.
• Audio files are in WAV format.
• Both Train and Test splits include a dedicated Noise class.

---

Classes

• Chord classes: Isolated guitar chords (e.g. A, Am, A#, etc.)
• Noise class:
  Contains various non-chord sounds such as:
  • silence
  • hand movement
  • environmental background noise

The Noise class is intended to improve model robustness and reduce false chord detections during pauses between chord changes.

---

Recording Protocol

Training Set Protocol

Each chord class in the training set contains 25 recordings, following a structured and diverse strumming and fingering protocol.

Fingering 1

1. Regular downstroke
2. Regular upstroke
3. Fast downstroke
4. Fast upstroke
5. Slow downstroke (string by string)
6. Slow upstroke (string by string)
7. ⬇️⬇️ ⬆️⬆️ ⬇️ ⬆️
8. ⬇️⬆️⬇️⬆️⬇️⬆️
9. ⬇️⬇️⬇️⬇️⬇️⬇️

Fingering 2

10–18. Same protocol as steps 1–9 with an alternative fingering

Fingering 3

19–25. Repeat steps 1–7 with another fingering variation

---

Test Set ProtocolThe test set uses a simpler and more compact protocol, focused on evaluating generalization.

1. Regular downstroke
2. ⬇️⬇️⬇️⬇️⬇️⬇️
3. Rapid ⬇️⬆️

Different fingering: 4. Slow downstroke ⬇️
5. ⬇️⬆️⬇️⬆️⬇️⬆️
6. Rhythmic ⬇️⬆️

---

Dataset Statistics

• Recordings per chord (train): 25
• Recordings per chord (test): 6
• Average duration per chord class recording: ~5.5 seconds
• Total audio per chord class: ~135 seconds
• Leading silence before chord onset: ~1 second
• Splits: Train / Test
• Includes Noise class: Yes

---

Intended Use

This dataset is suitable for:

• Guitar chord classification
• Real-time chord recognition systems
• Robust audio classification in noisy environments
• Music information retrieval (MIR) research

---

Limitations

• Recorded using a single guitar and player
• Acoustic guitar only
• Chord voicings reflect common fingerings but are not exhaustive

Future versions may include:

• synthetic data
• additional guitars
• extended chord vocabulary

---

License

This dataset is released under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.

You are free to use, modify, and redistribute the dataset, including for commercial purposes, provided appropriate credit is given.

---

Citation

If you use this dataset in academic research or commercial applications, please cite:

@dataset{isolated_guitar_chords,
  author = {Your Name},
  title = {Isolated Guitar Chords Dataset},
  year = {2025},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/USERNAME/isolated-guitar-chords}
}