Good candidates
Moving vocal harshness, guitar ringing, resonant room notes, cymbal spikes and occasional narrow mix buildup that changes over time.
Free · deterministic local scan · adaptive multiband DSP
Dynamic Resonance Suppressor
Choose a frequency window, run a repeatable local scan and review the strongest tonal hotspots before you process. Reduce moving whistles, ringing notes and harsh peaks, compare the removed Delta signal, then export a reviewed WAV.
Adaptive resonance map
Spectrum + live reduction
SpectrumReductionProfile
Load audio to wake the analyzerPeaks rise upward. Red reduction moves downward.
Text chart fallback
ProfileStrongest hotspots
- 01No profile
- 02No profile
- 03No profile
Input−∞ dBFS
Maximum reduction0.0 dB
Output−∞ dBFS
EngineWaiting
Rendering locally…
Ready. Your audio never needs to leave this device.
- Detection
- 8, 12 or 18 log-spaced band-pass followers compare current energy inside the selected frequency window.
- Processing
- Only excess local peaks receive attack/release-smoothed, depth-limited subtraction.
- Privacy
- Decode, live AudioWorklet monitoring and exact-frame WAV encoding happen on this device.
How to use adaptive resonance suppression
- Open a vocal, instrument, mix or dialogue file. Set Focus Low and Focus High around the region you want to inspect.
- Choose Linked stereo to preserve shared channel balance, or Independent when the problem is isolated to one channel, then run Learn.
- Review the three text hotspots, start the live monitor and raise Depth only until the distracting ring settles.
- Use Delta. If it sounds like the full performance rather than isolated ringing, back off Depth or raise Selectivity.
- Compare Processed and Original at the shared playhead, check output headroom, and export WAV.
How the adaptive DSP works
The engine runs a log-spaced bank of causal band-pass detectors inside the selected Focus window. For every band it compares the smoothed envelope with adjacent bands and a small broadband reference. A region becomes eligible only when its local prominence clears the Selectivity threshold. Depth caps the attenuation; Attack and Release smooth movement; Sharpness changes detector Q. Linked stereo shares the reduction decision, while Independent stereo follows each channel separately. The wet path subtracts only the bounded detected-band component, then Mix and Output are applied.
Learn is a deterministic scan, not a listening model: the same decoded audio, frequency window and settings produce the same review profile. It does not recognize instruments or artistic intent. The visualizer reports detector envelopes and requested attenuation, not a laboratory FFT measurement or perceptual loudness score.
When to use it — and when not to
Prefer static EQ
Use a normal equalizer when the same broad tonal imbalance is present throughout the entire recording and should stay corrected.
Prefer de-essing
Use a de-esser when the problem is specifically speech or vocal sibilance in a known high-frequency region.
Technical limits
- Very sharp filters, high Depth and short Attack can add phase character or reshape transients.
- The detector reacts to energy relationships; it does not recognize instruments, notes, words or artistic intent.
- Delta can sound louder or quieter than the processed path and is diagnostic, not a delivery signal.
- Output is sample-clamped for safety. The tool does not measure true peak or guarantee streaming loudness compliance.
- Live processed monitoring requires AudioWorklet. Local offline WAV export remains available when decoding succeeds.
Dynamic Resonance Suppressor: quick answer and technical limits
Quick answer: An adaptive multiband resonance suppressor that follows moving whistles, ringing notes and harsh peaks instead of applying one fixed equalizer cut.
- Best for
- Taming changing vocal harshness, guitar ringing, cymbal spikes, room modes and intermittent narrow buildup while preserving unrelated frequencies.
- How it works
- Eight, twelve or eighteen log-spaced band-pass followers compare each envelope with adjacent bands and a broadband reference; only locally prominent regions receive attack/release-smoothed, depth-limited subtraction.
- What you get
- Live Processed, Original and Delta audition, a spectrum-plus-reduction map, measured maximum gain reduction and an exact-frame local WAV export.
Know before you use it: This detector does not recognize instruments or artistic intent. High Depth, low Selectivity, very sharp bands or fast timing can hollow tone or reshape transients; Output is not a true-peak limiter.
Privacy: Selected audio is processed in this browser and is not uploaded to AudioWrench. Normal page assets can still be requested as described in the privacy notice. Privacy details →
FAQ
What is dynamic resonance suppression?
It is frequency-selective attenuation whose amount changes over time. This implementation compares log-spaced band energy with neighbouring bands and reduces only regions that rise above that local spectral baseline.
How is this different from a static equalizer?
A static equalizer applies the same curve throughout the recording. Adaptive suppression follows moving peaks and releases the cut when a peak no longer stands above nearby spectral energy.
What does Delta listen play?
Delta plays the component removed by the adaptive bands. It helps reveal whether the settings are catching a narrow resonance or taking useful tone and transients with it.
Can aggressive settings damage the sound?
Yes. High Depth, low Selectivity, very sharp bands or fast timing can hollow tone, emphasize phase character or make the processing audibly move. Compare at matched level and inspect Delta before export.
What does Learn do?
Learn runs a deterministic local scan inside the selected frequency window and reports the three strongest tonal hotspots. It creates a review profile; it does not identify instruments or make an artistic decision for you.
How do linked and independent stereo modes differ?
Linked uses one shared reduction decision for both channels to preserve stereo balance. Independent lets each channel respond to its own detector profile, which can reduce channel-specific ringing but may change the stereo image more.
Is my audio uploaded?
No. Selected audio is decoded, monitored, processed and encoded on this device. Normal page assets may still be requested according to the privacy notice.
Related frequency-control tools
Privacy: Selected audio is processed in this browser and is not uploaded to AudioWrench. Normal page assets can still be requested as described in the privacy notice.