Real-time detection of feedback, hum, polarity, noise, distortion, and comb filtering
Start measurement to detect feedback
Start measurement to detect hum
No comb filtering detected
Awaiting measurement
Awaiting measurement
Audio systems in live sound, installed AV, and recording environments suffer from a predictable set of problems that degrade intelligibility, introduce unwanted artefacts, and reduce headroom. Identifying these problems quickly is the difference between a successful event and an audience enduring painful feedback or incomprehensible speech. SonaVyx Problem Detector runs real-time analysis on your microphone input, using Rust-compiled WASM DSP routines, to pinpoint the most common faults in seconds rather than minutes.
Feedback occurs when a microphone picks up its own amplified signal from a loudspeaker, creating a regenerative loop. The result is a sustained, often ear-splitting tone at whichever frequency has the highest open-loop gain. The Problem Detector identifies these resonant peaks by comparing spectral energy to the broadband average. For each feedback-prone frequency, it suggests notch filter parameters (centre frequency, Q factor, and depth) that a system engineer can apply on a parametric EQ or DSP processor. Early detection of pre-feedback ringing can prevent full-blown howlround before it occurs.
The 50 Hz (Europe, Asia, Africa) or 60 Hz (Americas, Japan) mains frequency often couples into audio circuits through ground loops, poorly shielded cables, or proximity to transformers. Hum typically manifests as a fundamental plus strong odd and even harmonics (100/120 Hz, 150/180 Hz, 200/240 Hz, etc.). The hum detector analyses the harmonic series to determine whether you are dealing with 50 Hz or 60 Hz mains, and the harmonic pattern helps diagnose the cause: a dominant second harmonic suggests a full-wave rectifier issue, while strong odd harmonics point toward magnetic coupling from a transformer. Ground loops are resolved with isolation transformers, balanced connections, or ground lift adapters.
Reversed polarity (pin-2/pin-3 swap on an XLR cable, or a driver wired backwards) causes destructive interference when the affected channel is summed with others. A single reversed subwoofer in a distributed system can cancel more low-frequency energy than it adds. The polarity checker analyses the impulse response to determine whether the initial transient is positive-going (normal) or negative-going (reversed). Comb filtering, where two copies of a signal arrive with a small time offset, creates a characteristic pattern of alternating peaks and nulls in the frequency response. The detector estimates the delay offset and first null frequency so you can identify the physical cause, whether it is a reflection off a nearby surface, duplicate mic pickup, or an improperly aligned driver in a multi-way loudspeaker.
Every component in the signal chain contributes noise, from microphone self-noise to preamp hiss to DAC quantisation. The noise floor measurement gives you a single dBA figure and NC rating for the system at rest, allowing comparison against target levels for the application. Total Harmonic Distortion plus Noise (THD+N) measures the ratio of unwanted harmonic content and noise to the desired fundamental signal. Values below 0.1% are generally inaudible; above 1% indicates a fatigued amplifier, a clipping stage, or a damaged driver. The clipping detector counts samples at or near digital full scale, alerting you before audible distortion becomes damaging. Together, these measurements let you validate signal chain integrity from source to output.
The detector identifies feedback-prone frequencies before they reach full oscillation by detecting spectral peaks with high Q and rising gain. While SonaVyx does not directly control your mixer or DSP, it provides the exact notch filter parameters (frequency, Q, depth) needed to suppress each resonance. Applying these on your system processor eliminates the risk before feedback becomes audible.
The detector examines the 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. This works reliably even in environments with significant broadband noise because the harmonic structure is highly distinctive.
Comb filtering occurs when two copies of the same signal combine with a time offset, creating alternating peaks and nulls in the frequency response that resemble the teeth of a comb. Even a 1 ms delay creates nulls every 1000 Hz. In live sound, common causes include reflections from nearby surfaces, two microphones picking up the same source at different distances, or misaligned drivers in a loudspeaker cabinet. The result is a thin, hollow, or nasal sound quality.
Professional audio equipment typically achieves THD+N below 0.01% at rated output. Values below 0.1% are generally considered transparent. Between 0.1% and 1% distortion may become audible on sustained tones or critical program material. Above 1% usually indicates a problem such as a clipping amplifier stage, a damaged loudspeaker driver, or a faulty cable connection.
Browser-based echo cancellation, noise suppression, and automatic gain control algorithms are designed for voice communication. They actively modify the audio signal by removing tonal content (which would hide hum), compressing dynamics (which would mask clipping), and suppressing low-level signals (which would affect noise floor measurements). Disabling these features provides the raw microphone signal needed for accurate acoustic analysis.