Brumelet Spectral Edit

Reduce a whistle or tone in a short recording without lowering the whole take. Brumelet lets you select the exact time and frequency region, hear the result and inspect the sound that was removed.

A focused offline tool for your own recordings up to 12 seconds. One upfront purchase unlocks every feature on iPhone and iPad.

SELECT THE SOUND

Open a mono or stereo WAV, AIFF or M4A from Files. Draw directly on the actual spectrogram, or enter precise start/end seconds and low/high Hz. Add up to 8 regions, name them and choose attenuation from −6 to −60 dB. Enable, disable, delete or undo edits. The display shows channel 1 with time and frequency rulers.

HEAR WHAT CHANGED

Render your edits and compare Original and Processed at the same amplitude scale. Listen to Original, Processed or Removed. Removed is source minus processed audio, so you can check whether wanted sound was affected too. This is a manual frequency attenuation tool; it does not identify noise, regenerate missing audio or separate voices and instruments.

KEEP USEFUL FILES

Export a 9-file ZIP containing Processed.wav, Removed.wav, the unchanged source file, before/after spectrogram images, Regions.csv, an editable Project.json, a real proof PDF and a readme. Share the ZIP or individual files and preview the PDF inside the app. The unscaled Float32 WAVs preserve the source sample rate and channel count.

REOPEN YOUR WORK

Save up to 12 local projects with their original source audio and regions. Reopen, rename, duplicate or delete saved projects. Import an exported Project.json through Files to restore its audio and settings. Open work is retained across relaunch. iPad places the frequency canvas beside the inspector and supports landscape.

Brumelet starts empty. An optional Whistle study is synthesized on your device to help you learn the selection and comparison controls.

KNOW THE LIMITS

• WAV, AIFF and device-decodable M4A; mono or stereo; integer sample rates from 8–96 kHz.

• Up to 12 seconds, 1 million decoded frames and 24 MiB per audio source. Project JSON up to 34 MiB; saved sources up to 200 MiB in total.

• Regions use smooth time and frequency edges. Overlapping regions multiply attenuation. Processing windows can affect approximately 21 ms beyond a selected time edge. Wanted sound inside a region can change.

• Playback uses a common safety gain when needed for a −1 dBFS sample peak. Exported Float32 WAVs are unscaled and can exceed full scale; check your destination editor. This is not limiting or true-peak processing.

• No microphone recording, AUv3, live processing, resampling or automatic restoration. Keep originals separately and use audio you have rights to process and share.

No subscription, in-app purchase, account, ads, analytics or developer upload. Audio and projects stay on your device unless you choose to share them.