Manual

• Tap Microphone on the source screen to start the live RTA and spectrogram.
• Suspected howls are drawn as red boxes with their frequency, and the last 5 seconds accumulate in the side list.
• The last 5 minutes are recorded automatically; press Stop to send that recording into file analysis (“replay”).
• Change the input device and gain (dB) in ⚙️ Settings.

• Pick a file, then set a region. Drag the seekbar to select, and wheel/pinch to zoom down to milliseconds.
• Press Analyze region → to render the whole region as one spectrogram with howls boxed.
• Loop-play to listen and confirm.

• On the analysis screen, click/drag the graph to create a notch (cut one frequency). Right-click the spectrogram to set the frequency directly.
• It applies to playback in real time so you can hear whether the howl is gone.
• Use ✂ Download to export the region as MP3 (the notch frequency is in the filename).

At 0 nothing is detected; at 100 it catches as much as possible. Moving the slider re-computes the visible past region instantly, in both mic and analysis.

• Space — play/stop
• ← → — seek ±5s (file)
• Seekbar wheel/pinch — zoom

Mic input, files and recordings are processed entirely in your browser and never sent to a server. Settings, regions and notches are kept only in this browser’s localStorage.