Ambient noise measurement (decibel meter/sound level meter plotting tool) | ZeroTools

This system provides an advanced acoustic analysis environment that runs on a browser by integrally controlling the media stream audio source node and analyzer node, which are the core of the web audio application programming interface.

A continuous sound wave signal obtained from a microphone installed in the user's device undergoes analog-to-digital conversion and is expanded into memory as pulse code modulation data.

The total amount of acoustic energy in the time domain is quantified by applying the root mean square or effective value operation to this expanded waveform data array.

Specifically, call the get float time domain data method of the analyzer node to obtain minute amplitude fluctuations on the time axis as a floating point number array.

The effective value is calculated by squaring the amplitude value of each sample obtained, finding the sum, dividing it by the number of samples, and then taking the square root.

This effective value becomes a basic index representing the physical pressure of sound and functions as an absolute reference value in the subsequent decibel conversion process.

Because it uses an architecture that processes acoustic data in real time only within the browser's local memory, no audio data is sent to an external server, ensuring continuous measurement with complete privacy protection.

Furthermore, by optimizing the sampling rate and buffer size, we have built an advanced pipeline that minimizes processing delays and completes the time lag from microphone input to effective value calculation with millisecond precision that humans cannot perceive.

In order to adapt the calculated physical sound pressure effective value to the loudness of sound that humans actually feel, this system uses software to simulate the A-weighted frequency correction curve specified by the International Electrotechnical Commission.

Human hearing does not have uniform sensitivity; it has non-linear frequency characteristics, being most sensitive in the 3,000 to 4,000 Hz band, and extremely less sensitive in the bass range below 100 hertz and the treble range above 10,000 hertz.

If noise is evaluated based only on physical sound pressure without correcting this physiological hearing sensitivity, a fatal discrepancy will occur in which the energy in the low frequency range will be overestimated.

To solve this problem, we digitally apply a weighting filter based on an A-weighted transfer function to each bin's spectral data converted to the frequency domain.

Specifically, a specified amount of attenuation and amplification is added for each frequency component obtained by Fourier transform to derive the auditory corrected sound pressure along the equal loudness curve.

Only after this correction process is the simple sound pressure level sublimated into a noise evaluation index expressed in units of decibel A, or dBA, which is extremely close to that of the human ear.

This digital filtering process can also be achieved by cascading multiple stages of biquad filter nodes in the web audio application programming interface, but in this architecture, a more precise approximation of the A-characteristic curve is achieved by directly performing mathematical convolution operations on the frequency domain data obtained from the analyzer node.

The noise value converted to decibel A is compared in real time with the environmental sound standard database built into this system, and it is instantly determined to which physical situation the current acoustic state corresponds.

This judgment algorithm performs strict threshold control considering the characteristics of decibel values ​​expressed on a logarithmic scale.

A small fluctuation band around 20 decibels is defined as an extremely quiet environment that can be classified as the sound of leaves touching each other or whispering.

The 40 decibel band is recognized as the background noise level in a quiet residential area or inside a library, and serves as the optimal environmental reference value for tasks that require concentration.

If it exceeds 60 decibels, it is judged as normal conversation or environmental noise in an office, and is treated as a standard daily noise level that does not interfere with communication.

If the threshold of 80 decibels is exceeded, the environment is considered to be comparable to a moving train or a busy intersection, and a warning flag for long-term exposure is raised internally.

Furthermore, if it detects an extremely high sound pressure band of 100 decibels, equivalent to the level found under railway guards or near rock jackhammers, it will immediately issue a signal exceeding the limit value.

At the same time, the system runs a peak sound pressure detection algorithm that detects sudden impact sounds from continuous sound pressure fluctuations.

This is a mechanism that tracks the maximum absolute value of time domain data frame by frame and sets a certain hold time to accurately capture the energy of instantaneous plosive sounds and impact sounds, and records it as a peak decibel value that is independent of the average decibel value.

We implement a real-time frequency analysis mechanism that utilizes a fast Fourier transform algorithm running inside the analyzer node to visualize not only the overall amount of sound pressure but also the distribution of frequency components that make up the sound.

The frequency-domain array data extracted through the GetBite Frequency Data method is the result of converting a time-domain waveform into a frequency-domain spectrum, detailing which pitches and intensities are present.

The size of the fast Fourier transform is set to an optimal power-of-two value considering the trade-off between resolution and processing speed, and the band up to the Nyquist frequency is divided into equally spaced bins for analysis.

The resulting spectral data is rendered smoothly in the browser's drawing area at 60 frames per second using a canvas application programming interface.

The amount of energy in each frequency bin is mapped as a bar graph with height or as a continuous line graph, and the energy distribution from bass to treble is dynamically expressed as waving.

This drawing process uses a recursive loop of requested animation frames to avoid blocking the main thread while providing extremely low-latency visual feedback.

Furthermore, continuous measurement results are buffered and the average decibel value changes over time and the maximum decibel value during the observation period are recorded as a log and superimposed on the graph as a trend line, making it possible to intuitively grasp not only momentary noise but also long-term fluctuation trends in the acoustic environment.

The decibel meter display mechanism, which forms the core of the user interface, is controlled by advanced drawing logic that visualizes the calculated dBA value as an analog instrument panel or high-definition digital indicator.

An easing function is applied to the movement of the meter needle and update of the indicator value to make logarithmic scale fluctuations look natural, ensuring accurate response while maintaining visual smoothness even in the face of sudden changes in sound pressure.

This meter display does more than simply display numerical values; it also includes a dynamic color grading system linked to the aforementioned criteria.

At safe ambient sound levels, the meter maintains a cool color display, but when a certain noise level is exceeded, the meter gradually transitions to a warm color to alert the user visually.

Additionally, an auditory risk warning system based on occupational safety and health standards and World Health Organization guidelines is integrated.

If you are continuously exposed to an environment exceeding 80 decibels for more than a certain period of time, or if a momentary peak sound pressure exceeding 100 decibels is detected, a visual meter will flash red and a warning message recommending hearing protection will be forcibly overlaid on the screen.

This risk assessment is based on a cumulative acoustic energy calculation that considers a time-weighted average rather than a single instantaneous value, creating an active feedback loop to accurately detect environments that increase the risk of acoustic hearing loss and protect the user's health.

The highly accurate acoustic analysis function provided by this system goes beyond a simple sound level meter and has an architecture that can be applied to a variety of specialized applications.

When continuously measuring environmental noise in living or working environments, a log recording mechanism that operates in the background accumulates long-term measurement data, which can be used as basic data for environmental assessments that quantitatively evaluate noise trends and average quietness during specific time periods.

It is also extremely effective as a quiet environment check tool in recording studios and when recording podcasts. By checking the microphone's directivity and ease of picking up environmental noise depending on its installation location using real-time spectrum display and A-weighting corrected decibel values, you can powerfully support building the optimal recording environment.

It can also be applied to signal level evaluation of instruments and line inputs input through audio interfaces. Delay-free level meters and frequency distribution visualization are essential indicators in audio engineering, whether it's monitoring peak levels just before clipping distortion occurs or adjusting the volume balance between multiple input sources.

Complete with a web browser, this system provides instant professional sound measurement and level adjustment on any platform without the need for specialized hardware equipment, serving as a comprehensive solution for seamlessly optimizing audio input quality and assessing environmental sound.