🎵 Artificial Grammar-Based Chord Sequence Generator

This repository contains a Hidden Markov Model (HMM)-based Python tool for generating musically informed chord and pitch sequences. It simulates probabilistic musical grammar, with an emphasis on realistic transitions, constrained pitch repetition, and harmonically meaningful sequences — suitable for auditory neuroscience, BCI stimulus creation, or music cognition experiments.

---

🧠 Purpose

To generate synthetic chord sequences governed by a structured probabilistic grammar. These sequences can be used as stimuli for behavioral or neuroimaging studies investigating musical expectation, prediction, or perception.

---

🚀 Features

• Hidden Markov model (HMM)–based state transitions between chords
• Transition matrix tailored to reflect harmonic structure
• Constrained pitch selection with probabilistic weighting
• Repetition-avoidance logic for pitch realism
• Output in both chord sequences and flattened pitch sequences
• Probabilistic weighting favors cadences and closure on final chords

---

📜 Script Overview

1st_grammar_seq_generator.py

This script includes:

• generate_chord_sequence(length):
  Constructs a sequence of chords based on an HMM-like transition matrix.

• pitch_from_chord_sequence(chord_sequence):
  Samples pitches from each chord with weighted probabilities and repetition suppression.

• main():
  Generates a batch of sequences and stores them in memory. Configurable for experimental batch runs.

---

💡 Future Directions

• MIDI/audio export of sequences
• Integration with timbre synthesis engines
• Visualization of transition probabilities
• Extension to polyphonic or rhythmic structures

📖 Citation

If you use this tool in your own work, please cite:

spraveena. (2024). spraveena/ArtificialGrammarGenerator: Hidden Markov Model (HMM) for
Chord Sequence Generation (v1.0.0). Zenodo. https://doi.org/10.5281/zenodo.13937536

the Bibtex format is as follows:

@software{spraveena_2024_13937536,
  author       = {spraveena},
  title        = {spraveena/ArtificialGrammarGenerator: Hidden
                   Markov Model (HMM) for Chord Sequence Generation
                  },
  month        = oct,
  year         = 2024,
  publisher    = {Zenodo},
  version      = {v1.0.0},
  doi          = {10.5281/zenodo.13937536},
  url          = {https://doi.org/10.5281/zenodo.13937536},
}