tabsynth

Convert pre-detected musical events (notes and chords) into playable guitar tablature via candidate generation and dynamic programming optimization.

Overview

tabsynth is a production-ready Python package that transforms musical event data into optimized guitar tablature. It focuses on the synthesis and optimization of tablature from already-detected musical events—audio detection is out of scope.

Key Features

• Event-Based Input: Accepts NoteEvent (single pitch) and ChordEvent (multiple pitches) objects
• Candidate Generation: Creates multiple playable fingering options for each event
• DP Optimization: Uses Viterbi-style dynamic programming to find the optimal sequence minimizing hand movement and difficulty
• Chord Template Matching: Built-in library of common chord shapes (open chords, barre chords)
• Multiple Output Formats: ASCII tablature, JSON, and compact text representations
• Extensible: Easy to add custom chord templates

Installation

pip install tabsynth

Or install from source:

cd tabsynth
pip install -e .

For development with tests:

pip install -e ".[dev]"

Quick Start

Basic Usage

from tabsynth import NoteEvent, ChordEvent, events_to_tablature

# Define some musical events
events = [
    NoteEvent(pitch_hz=329.63, start=0.0, duration=0.5),  # E4
    NoteEvent(pitch_hz=369.99, start=0.5, duration=0.5),  # F#4
    NoteEvent(pitch_hz=392.00, start=1.0, duration=0.5),  # G4
    NoteEvent(pitch_hz=440.00, start=1.5, duration=0.5),  # A4
]

# Convert to tablature
tablature = events_to_tablature(events, output_format="ascii")
print(tablature)

Working with Chords

from tabsynth import ChordEvent, events_to_tablature

# E major chord
chord = ChordEvent(
    pitches_hz=[329.63, 415.30, 493.88],  # E, G#, B
    start=0.0,
    duration=1.0
)

tablature = events_to_tablature([chord], output_format="ascii")
print(tablature)

Using the Pipeline API

from tabsynth import TabSynthPipeline, NoteEvent

# Create a pipeline with custom settings
pipeline = TabSynthPipeline(
    max_fret=12,
    tolerance_cents=50.0
)

# Process events
events = [NoteEvent(pitch_hz=440.0, start=0.0, duration=0.5)]
result = pipeline.process(events, output_format="json")

Command-Line Interface

The package includes a demo CLI:

# Show help
tabsynth help

# Run demos
tabsynth demo-notes
tabsynth demo-chords
tabsynth demo-mixed

Architecture

Data Flow

Events (Notes/Chords)
    ↓
Candidate Generation (multiple fingerings per event)
    ↓
Cost Calculation (hand position, stretch, barre difficulty)
    ↓
DP Optimization (Viterbi algorithm)
    ↓
Rendering (ASCII/JSON/Compact)

Core Components

1. model.py: Data structures for events and playable states
2. fretboard.py: Pitch helpers and fretboard modeling
3. templates.py: Chord shape library
4. candidates.py: Generate playable fingering candidates
5. cost.py: Cost functions for transitions and states
6. optimize.py: Dynamic programming optimizer
7. render.py: Output formatting
8. pipeline.py: End-to-end integration

Extending with Custom Chord Templates

You can add your own chord shapes:

from tabsynth import ChordTemplate, TabSynthPipeline

# Define a custom chord
my_chord = ChordTemplate(
    id="D_major_open",
    frets=[None, None, 0, 2, 3, 2],  # Strings 6,5,4,3,2,1
    barre=False,
    pitch_classes={"D", "F#", "A"},
    span=3,
    tags={"open", "basic"}
)

# Use it in a pipeline
pipeline = TabSynthPipeline(templates=[my_chord])

Template Fields

• id: Unique identifier
• frets: List of 6 fret numbers or None (for strings 6→1)
• barre: Whether the chord requires barre technique
• pitch_classes: Set of note names in the chord
• span: Fret span (max - min among fretted notes)
• tags: Descriptive tags (e.g., "open", "barre", "E-shape")

Output Formats

ASCII Tablature

Guitar Tablature
========================================

e|0-5-7-5|
B|0-7-8-7|
G|1-6-7-6|
D|2-7-9-7|
A|2-7-9-7|
E|0-5-7-5|

Chords:
  E_major_open

JSON

[
  {
    "index": 0,
    "start": 0.0,
    "duration": 0.5,
    "kind": "note",
    "strings": [1],
    "frets": {"1": 5},
    "mean_fret": 5.0,
    "min_fret": 5,
    "max_fret": 5,
    "requires_barre": false,
    "chord_id": null
  }
]

Compact

0: t=0.00 note [1:5]
1: t=0.50 note [1:7]
2: t=1.00 E_major_open [6:0,5:2,4:2,3:1,2:0,1:0]

API Reference

Events

NoteEvent(pitch_hz: float, start: float, duration: float, confidence: float = 1.0)
ChordEvent(pitches_hz: list[float], start: float, duration: float, confidence: float = 1.0)

Pipeline

events_to_tablature(
    events: list[Event],
    output_format: "ascii" | "json" | "compact" = "ascii",
    templates: list[ChordTemplate] = None,
    max_fret: int = 15,
    tolerance_cents: float = 50.0
) -> str

Optimization

optimize_sequence(
    events: list[Event],
    templates: list[ChordTemplate] = None,
    max_fret: int = 15,
    tolerance_cents: float = 50.0
) -> list[PlayableState]

Testing

Run tests with pytest:

pytest tabsynth/tests/

Tests cover:

• Fretboard and pitch calculations
• Candidate generation
• Optimization correctness
• End-to-end pipeline

Design Principles

1. Minimal Dependencies: Core functionality requires only Python 3.10+
2. Typed: Full type hints for better IDE support
3. Extensible: Easy to add new chord templates and cost functions
4. Focused: Audio detection is out of scope; assumes pre-detected events
5. Optimized: DP algorithm ensures efficient fingering sequences

Limitations

• Audio detection not included: Requires pre-detected pitch events
• Standard tuning focus: Primary support for EADGBE tuning
• Template-based chords: Chord recognition relies on predefined templates
• No timing quantization: Preserves input event timing as-is

Contributing

To add new chord templates to the library:

1. Define a ChordTemplate with accurate fret positions
2. Specify pitch classes and tags
3. Add to V1_TEMPLATES in templates.py
4. Write tests to verify matching

License

MIT License - see LICENSE file for details.

Version

Current version: 0.1.0