noisemonitor

A python package for sound level data analysis.

See the Usage Guide for detailed examples and the API Reference for function documentation.

⚠️ Version 1.0.0 - Breaking Changes: This major update introduces a new functional API and is not backward-compatible with previous versions. See the Migration Guide for details.

Table of Contents

• Key Features
• Installation
• Quick Start
• Core Modules
• Citation
• License

Key Features

• Acoustic indicators - Implements standard (Leq, Lden, L90, etc.) and research-based noise indicators (HARMONICA, Number of Noise Events, etc.)
• Flexible data loading - Support for CSV, Excel, and TXT formats with automatic datetime parsing
• Two analysis modes - Summary indicators (discrete values) and profile indicators (time series)
• Multiprocessing support - Fast processing of large datasets
• Easy visualization - Built-in plotting functions with customizable styles
• Data coverage validation - Automated quality checks with configurable thresholds
• Weather integration - [Canada only] Merge and analyze data with Canadian weather station data

Installation

Using PyPI

pip install noisemonitor

For weather integration (Canada only)

pip install noisemonitor[weather]

If you want to install the latest development version:

pip install git+https://github.com/valerianF/noisemonitor

Using Anaconda

Anaconda users can install using conda-forge:

conda install -c conda-forge noisemonitor

Quick Start

noisemonitor is designed for flexibility and ease of use. All analysis functions accept an optional column parameter, allowing you to specify which data column to analyze (e.g., to work with datasets containing multiple sound level measurements or frequency bands). Most functions return results as pandas DataFrames for easy manipulation.

import noisemonitor as nm

# Load data
df_1m = nm.load(
    'tests/data/test_data_laeq1m.csv',
    datetimeindex=0, # Column index for datetime
    valueindexes=1,  # Column index(es) for sound levels
    header=0,        # Header row index
    sep=','
)

# Compute Lden
lden = nm.summary.lden(df_1m)
lden.head()

lden	lday	levening	lnight
56.1	51.75	50.08	49.23

# Compute weekly profiles
weekday_profile = nm.profile.periodic(
    df_1m,
    hour1=23,
    hour2=22,
    day1='monday',
    day2='friday',
    win=3600,    # 1-hour window
    step=1200    # 20-minute step
)

weekend_profile = nm.profile.periodic(
    df_1m,
    hour1=23,
    hour2=22,
    day1='saturday',
    day2='sunday',
    win=3600,
    step=1200
)

# Visualize
nm.display.compare(
    [weekday_profile, weekend_profile],
    ['Weekday', 'Weekend'],
    'Leq',
    fill_background=True,
    title='Weekly Noise Profiles'
)

# Computa HARMONICA indexes
harmonica = nm.summary.harmonica_periodic(df_1s) # requires 1s resolution data
nm.display.harmonica(harmonica) # Visualize

Core Modules

nm.load()

Load data from CSV, Excel, or TXT files (or list of files) with automatic datetime parsing.

df = nm.load('data.csv', datetimeindex=0, valueindexes=1, header=0, sep=',')

nm.filter

Filter data by datetime, remove outliers, or filter by weather conditions.

df_filtered = nm.filter.extreme_values(df, min_value=30, max_value=100)

nm.summary

Compute discrete sound level indicators: Leq, Lden, HARMONICA, frequency analysis, histograms, etc.

overall_lden = nm.summary.lden(df)
daily_lden = nm.summary.periodic(df, freq='D')
harmonica = nm.summary.harmonica_periodic(df)
freq_summary = nm.summary.freq_periodic(df_freq, freq='D')

nm.profile

Compute time-varying sound level profiles: time series, daily/weekly patterns, number of noise events, etc.

time_series = nm.profile.series(df, win=3600, step=1200)
weekday_profile = nm.profile.periodic(df, hour1=0, hour2=23, day1='monday', day2='friday', win=3600)
nne_profile = nm.profile.nne(df, hour1=0, hour2=23, background_type='L50', exceedance=5)

nm.display

Visualize results with line plots, heatmaps, and more.

nm.display.line(time_series, 'Leq', 'L10', 'L90', title='Noise Levels')
nm.display.compare([weekday_profile, weekend_profile], ['Weekdays', 'Weekend'], 'Leq')
nm.display.freq_map(freq_data['Leq'], title='Frequency Heatmap')

nm.weather (Canada only)

Integrate Environment Canada weather data to analyze weather impact on noise.

stations = nm.weather.weathercan.get_historical_stations(coordinates=[45.5, -73.6], radius=25)
df_weather = await nm.weather.weathercan.merge_weather(df, station_id=30165, wind_speed_flag=18)
contingency = nm.weather.weathercan.contingency_weather_flags(df_weather)

Citation

If you use noisemonitor, please consider citing us:

@inproceedings{fraisse2023noisemonitor,
    title={noisemonitor: A Python Package For Sound Level Monitor Analysis},
    author={Fraisse, Valérian},
    booktitle={Acoustics Week in Canada},
    year={2023}
}

Dependencies

• NumPy (numpy.org)
• pandas (pandas.pydata.org)
• Matplotlib (matplotlib.org)

License

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.