SpecPlotter

A Python package for creating spectrograms that looks good. Its parameters are carefully tuned by David Harwath to resemble the appearance of physically produced spectrograms, e.g., Kay Sona-Graph.

pip install specplotter

The easiest way to use SpecPlotter is via the command line:

# Basic spectrogram (PNG)
specplotter audio.wav -o output.png

# Basic spectrogram (PDF)
specplotter audio.wav -o output.pdf

# Full analysis with all components
specplotter audio.wav -o output.pdf --zcr --total-energy --lowfreq-energy --waveform

# Narrowband mode
specplotter audio.wav -o output.png --mode narrowband

from specplotter import SpecPlotter
import librosa

# Load an audio file
signal, sr = librosa.load('audio.wav', sr=16000)

# Create a SpecPlotter instance (default: wideband mode)
plotter = SpecPlotter()

# Plot spectrogram only (default behavior)
plotter.plot_spectrogram(signal)

# Or save to file
plotter.plot_spectrogram(signal, outfile='spectrogram.png')

from specplotter import SpecPlotter
import librosa

signal, sr = librosa.load('audio.wav', sr=16000)
plotter = SpecPlotter()

# Default: spectrogram only
plotter.plot_spectrogram(signal)

# Wideband mode (default)
plotter_wide = SpecPlotter(mode='wideband')

# Narrowband mode
plotter_narrow = SpecPlotter(mode='narrowband')

# Plot with different modes
plotter_wide.plot_spectrogram(signal)
plotter_narrow.plot_spectrogram(signal)

# Plot with all components
plotter.plot(
    signal,
    show_zcr=True,
    show_total_energy=True,
    show_lowfreq_energy=True,
    show_waveform=True
)

# Just spectrogram and waveform
plotter.plot(signal, show_waveform=True)

# Spectrogram with zero crossing rate
plotter.plot(signal, show_zcr=True)

import matplotlib.pyplot as plt

# Single axis (spectrogram only)
fig, ax = plt.subplots()
plotter.plot_spectrogram(signal, ax=ax)

# List of axes matching number of plots
fig, axes = plt.subplots(3, 1)  # For zcr, spectrogram, waveform
plotter.plot(
    signal,
    ax=list(axes),
    show_zcr=True,
    show_waveform=True
)

# Full plot with all components
fig, axes = plt.subplots(5, 1)
plotter.plot(
    signal,
    ax=list(axes),
    show_zcr=True,
    show_total_energy=True,
    show_lowfreq_energy=True,
    show_waveform=True
)

Note: When providing a list of axes, they must be in this order:

1. zcr (if show_zcr=True)
2. total_energy (if show_total_energy=True)
3. lowfreq_energy (if show_lowfreq_energy=True)
4. spectrogram (always included)
5. waveform (if show_waveform=True)

# Compute all features
features = plotter.compute_spectrogram(signal)

# Access individual features
processed_signal = features['processed_signal']
spectrogram = features['spectrogram']
zcr = features['zcr']
total_energy = features['total_energy']
lowfreq_energy = features['lowfreq_energy']

# Use features independently
import matplotlib.pyplot as plt
fig, ax = plt.subplots()
plotter._plot_spectrogram_on_axis(spectrogram, len(signal), ax=ax)

features = plotter.compute_spectrogram(signal)

# Plot individual components
fig, ax = plt.subplots()
plotter.plot_zcr(features['zcr'], ax=ax)

fig, ax = plt.subplots()
plotter.plot_total_energy(features['total_energy'], ax=ax)

fig, ax = plt.subplots()
plotter.plot_lowfreq_energy(features['lowfreq_energy'], ax=ax)

fig, ax = plt.subplots()
plotter.plot_waveform(features['processed_signal'], ax=ax)

The mode parameter sets default window sizes for wideband or narrowband analysis:

• Wideband (default): window_size=0.004, window_stride=0.001
• Narrowband: window_size=0.025, window_stride=0.01

All parameters can be customized when creating a SpecPlotter instance:

plotter = SpecPlotter(
    mode='wideband',              # 'wideband' or 'narrowband'
    sample_rate=16000,            # Sample rate (Hz)
    fnotch=60,                    # Notch filter frequency (Hz)
    notchQ=30,                    # Notch filter Q factor
    preemphasis_coeff=0.97,       # Pre-emphasis coefficient
    window_size=0.004,            # Window size (seconds), optional
    window_stride=0.001,          # Window stride (seconds), optional
    n_fft=1024,                   # Number of FFT points
    window=scipy.signal.windows.hamming,  # Window function
    db_spread=60,                 # Dynamic range (dB)
    db_cutoff=3,                  # Minimum dB value
    fig_height=10,                # Figure height (inches)
    inches_per_sec=10,            # Horizontal scaling
    zcr_smoothing_std=6,          # ZCR smoothing std dev
    zcr_smoothing_size=41,       # ZCR smoothing kernel size
    lowfreq_min=125,              # Low freq energy min (Hz)
    lowfreq_max=750,              # Low freq energy max (Hz)
)

You can override mode defaults by explicitly providing window_size and window_stride:

# Use wideband mode but with custom window settings
plotter = SpecPlotter(
    mode='wideband',
    window_size=0.005,   # Override default
    window_stride=0.002  # Override default
)

SpecPlotter includes a command-line interface for quick spectrogram generation.

specplotter <input_file> -o <output_file>

The output format (PNG or PDF) is automatically determined from the file extension.

Required:

• input_file: Path to input WAV file
• -o, --output: Output file path (must have .png or .pdf extension)

Analysis Mode:

• --mode {wideband,narrowband}: Analysis mode (default: wideband)

Additional Plots:

• --all: Show all additional plots (zcr, total-energy, lowfreq-energy, waveform)
• --zcr: Show zero crossing rate plot
• --total-energy: Show total energy plot
• --lowfreq-energy: Show low frequency energy plot
• --waveform: Show waveform plot

Audio Settings:

• --sample-rate FLOAT: Sample rate in Hz (default: 16000)

Processing Parameters:

• --fnotch FLOAT: Notch filter frequency in Hz (default: 60)
• --notch-q FLOAT: Notch filter Q factor (default: 30)
• --db-spread FLOAT: Dynamic range in dB (default: 60)
• --db-cutoff FLOAT: Minimum dB value to display (default: 3)

# Basic spectrogram
specplotter audio.wav -o spectrogram.png

# Full analysis (all components)
specplotter audio.wav -o analysis.pdf --all

# Narrowband mode with custom settings
specplotter audio.wav -o output.pdf --mode narrowband --db-spread 80

# Custom sample rate
specplotter audio.wav -o output.png --sample-rate 22050

# European line noise (50 Hz instead of 60 Hz)
specplotter audio.wav -o output.pdf --fnotch 50

For full help and all options:

specplotter --help

Initialize SpecPlotter with configurable parameters. See Configuration section above for all parameters.

Compute spectrogram and related features.

Parameters:

• signal (np.ndarray): Input audio signal

Returns:

• dict: Dictionary containing:
  • 'processed_signal': Preprocessed signal
  • 'spectrogram': Clipped log spectrogram
  • 'zcr': Zero crossing rate
  • 'total_energy': Total energy envelope
  • 'lowfreq_energy': Low frequency energy envelope

Plot spectrogram with optional additional features.

Parameters:

• signal (np.ndarray): Input audio signal
• ax (Axes or list of Axes, optional): Matplotlib axes to plot on
• show_zcr (bool): Whether to show zero crossing rate plot
• show_total_energy (bool): Whether to show total energy plot
• show_lowfreq_energy (bool): Whether to show low frequency energy plot
• show_waveform (bool): Whether to show waveform plot
• outfile (str, optional): If provided, save figure to file
• **kwargs: Additional keyword arguments passed to plotting functions

Returns:

• tuple: (figure, axes_dict)

Plot spectrogram only (convenience method).

Parameters:

• signal (np.ndarray): Input audio signal
• ax (Axes, optional): Matplotlib axes to plot on
• outfile (str, optional): If provided, save figure to file
• **kwargs: Additional keyword arguments

Returns:

• tuple: (figure, axes)

• Python >= 3.7
• numpy >= 1.19.0
• scipy >= 1.5.0
• librosa >= 0.8.0
• matplotlib >= 3.3.0

MIT License

This package uses python-semantic-release for automatic version management based on Conventional Commits.

The version is automatically determined from your commit messages:

• fix: or fix(scope): → patch version bump (0.0.1 → 0.0.2)
• feat: or feat(scope): → minor version bump (0.0.1 → 0.1.0)
• feat!: or fix!: or BREAKING CHANGE: → major version bump (0.0.1 → 1.0.0)

Simply push commits with conventional commit messages to the main branch:

git commit -m "feat: add new spectrogram visualization feature"
git push origin main

The GitHub Actions workflow will:

1. Analyze your commits since the last release
2. Determine the appropriate version bump
3. Update version numbers in pyproject.toml and __init__.py
4. Create a git tag
5. Generate/update CHANGELOG.md
6. Create a GitHub release
7. Build and publish to PyPI (if PYPI_API_TOKEN is configured)

Follow the Conventional Commits specification:

<type>[optional scope]: <description>

[optional body]

[optional footer(s)]

Types:

• feat: A new feature
• fix: A bug fix
• docs: Documentation only changes
• style: Code style changes (formatting, etc.)
• refactor: Code refactoring
• perf: Performance improvements
• test: Adding or updating tests
• chore: Maintenance tasks

Examples:

git commit -m "feat: add support for custom color maps"
git commit -m "fix: correct frequency calculation in spectrogram"
git commit -m "feat!: change API for plot_spectrogram method"
git commit -m "docs: update installation instructions"

Run tests using pytest:

# Install development dependencies
pip install -e ".[dev]"

# Run all tests
pytest

# Run with verbose output
pytest -v

# Run specific test file
pytest tests/test_specplotter.py

The test suite includes:

• Unit tests for SpecPlotter initialization and configuration
• Tests for spectrogram computation
• Tests for plotting functionality
• Tests for CLI interface
• Tests for error handling

Contributions are welcome! Please feel free to submit a Pull Request.

This project uses pre-commit hooks to ensure code quality. To set them up:

# Install pre-commit
pip install pre-commit

# Install the git hooks
pre-commit install

# Or run manually
pre-commit run --all-files

The pre-commit hooks will automatically:

• Remove trailing whitespace
• Fix end-of-file issues
• Check YAML, JSON, and TOML syntax
• Format code with ruff
• Run ruff linting (with auto-fix)
• Check for merge conflicts

1. Run the test suite: pytest
2. Run pre-commit hooks: pre-commit run --all-files
3. Ensure all tests pass
4. Ensure code formatting is correct