Back

Audio Problem Detector

Real-time detection of feedback, hum, polarity, noise, clipping, and comb filtering

AI Diagnostic

Problem Radar

Problem Priority

No problems detected

Feedback Detection

Start detection to find feedback

Hum Detector (50/60 Hz)

Start detection to check for hum

Comb Filter Detection

No comb filtering detected

Polarity Checker

Awaiting detection

Noise Floor

---dBA
NC Rating:NC-0

Clipping Detector

No clipping

Real-Time Audio Problem Detection for Sound Engineers

The SonaVyx Audio Problem Detector is a real-time analysis tool that continuously monitors your microphone input and identifies the most common faults in audio systems. Built on Rust-compiled WebAssembly DSP routines, it performs frequency-domain and time-domain analysis at 48 kHz with an 8192-point FFT, delivering professional-grade problem detection directly in your browser. No software installation, no dedicated hardware -- just a microphone and a web browser over HTTPS.

The Problem Radar

The circular radar display at the center of the interface arranges seven problem categories around a ring: Feedback, Hum, Polarity, Noise, Clipping, Comb Filter, and Phase. Each category is represented as a point on the radar, with its distance from the center indicating severity. When the system is clean, all points cluster near the center (green zone). As problems emerge, affected categories push outward through warning (yellow), problem (orange), and critical (red) zones. The filled polygon connecting all seven points gives you an instant visual fingerprint of your system's health -- a perfect circle at the center means everything is clean, while an irregular shape with spikes tells you exactly which problems need attention first.

Feedback Detection and Suppression

Acoustic feedback occurs when a microphone captures amplified sound from a loudspeaker, creating a regenerative loop that manifests as a sustained, often painfully loud tone. The Problem Detector identifies feedback-prone frequencies by analyzing spectral peaks that exhibit high Q factor (narrow bandwidth) and gain exceeding the broadband average. For each detected frequency, it provides the exact notch filter parameters needed to suppress the resonance: center frequency in Hz, Q factor, and recommended cut depth in decibels. These values can be entered directly into your system processor or mixing console's parametric EQ. Importantly, the detector can identify pre-feedback ringing -- the characteristic narrowing of a spectral peak just before it breaks into full oscillation -- allowing you to apply corrective filters before the audience hears anything.

Mains Hum and Ground Loop Diagnosis

Mains-frequency hum is one of the most common problems in audio systems, caused by ground loops, poorly shielded cables, proximity to power transformers, or faulty power supplies. The detector examines both the 50 Hz (Europe, Asia, most of Africa) and 60 Hz (Americas, parts of Asia) harmonic series simultaneously, identifying which fundamental is present and analyzing the harmonic pattern to suggest a probable cause. A dominant second harmonic (100 or 120 Hz) suggests full-wave rectifier ripple in a power supply. Strong odd harmonics point toward magnetic coupling from a nearby transformer. The harmonics bar chart displays the energy at each harmonic frequency, giving you a visual signature of the hum that helps diagnose whether the problem is electrical (ground loop) or magnetic (transformer proximity).

Polarity, Comb Filtering, and Phase Issues

Reversed polarity -- a wiring error that swaps the positive and negative conductors, typically pin 2 and pin 3 on an XLR connector -- causes destructive interference when the affected signal is summed with correctly wired channels. The polarity checker analyzes the impulse response to determine whether each channel's initial transient is positive-going (normal) or negative-going (reversed). Comb filtering results from two copies of the same signal arriving at the microphone with a small time offset, creating alternating peaks and nulls in the frequency response. The detector estimates the delay offset and first null frequency, which lets you calculate the physical path-length difference causing the problem. Even a 1 ms offset creates nulls every 1000 Hz, producing a thin, hollow sound quality that is immediately recognizable once you know what to listen for.

Noise Floor and Clipping Monitoring

The noise floor panel displays the system's broadband noise level in dBA and provides an NC (Noise Criteria) rating, allowing comparison against target levels for different applications -- NC-25 for recording studios, NC-30 for concert halls, NC-35 for conference rooms. The clipping detector counts the number of samples per second that reach or exceed digital full scale, alerting you to gain staging problems before they cause audible distortion or damage to loudspeaker drivers. Both measurements run continuously while detection is active, giving you real-time feedback as you adjust gain structure, add processing, or troubleshoot signal chain issues.

Frequently Asked Questions

Can the Problem Detector prevent feedback before it happens?
The detector identifies pre-feedback ringing -- spectral peaks with high Q factor and rising gain -- before they break into full oscillation. It provides exact notch filter parameters (frequency, Q, cut depth) that you can apply on your system processor to suppress each resonance before it becomes audible to the audience.
How does the hum detector distinguish between 50 Hz and 60 Hz mains?
It examines energy at both 50 Hz and 60 Hz along with their respective harmonic series (100/150/200 Hz vs 120/180/240 Hz). The fundamental with the stronger harmonic ladder is identified as the mains frequency. The harmonic pattern also helps diagnose the cause: dominant second harmonic suggests power supply ripple, while strong odd harmonics point to magnetic coupling.
What does the comb filter detector tell me?
It reports the estimated time delay between the two signal copies (in milliseconds), the frequency of the first null, and the null depth in decibels. From the delay, you can calculate the physical path-length difference (delay in seconds multiplied by 343 m/s for the speed of sound) to identify whether the cause is a surface reflection, duplicate microphone pickup, or misaligned driver.
What clipping rate is considered acceptable?
Ideally, there should be zero clips per second. Even occasional clipping (1-2 clips/sec) indicates that the signal is too hot at some point in the chain. Sustained clipping above 10 clips/sec causes audible distortion and risks damaging loudspeaker drivers, especially tweeters. The detector alerts you immediately so you can reduce gain before damage occurs.
Why should I disable browser audio processing for measurements?
Browser echo cancellation, noise suppression, and automatic gain control algorithms are designed for voice communication, not acoustic measurement. They actively modify the audio signal -- removing tonal content (hiding hum), compressing dynamics (masking clipping), and suppressing low-level signals (affecting noise floor readings). The Problem Detector requests raw microphone input with these features disabled for accurate analysis.