Introduction
Most noisy recordings have the same problem: a steady sound under everything, from a laptop fan, an air conditioner, a cheap microphone's hiss or electrical hum. It is quiet, but it makes speech harder to follow and makes a recording sound amateur.
ToolBuddy's Noise Remover reduces that steady noise in your browser. This guide explains how it works, which settings to use and how to check the result.
Know your noise
Noise reduction works on noise that stays the same over time. It learns that sound from the quietest moments of the recording, where only background remains, and turns it down everywhere.
Sudden sounds, such as a door, a cough, a dog or traffic going past, change too quickly to be learned as background and are mostly left alone. Music or voices in the background are not noise to the algorithm either.
- Works well: hiss, hum, fans, air conditioning, computer noise, room tone.
- Works poorly: sudden sounds, background music, other voices, echo.
Load and measure
Open the Noise Remover and drop a file on the selector, or tap Choose audio file. It accepts MP3, WAV, M4A, AAC, OGG/Opus, FLAC and WebM audio up to 500 MB (100 MB on phones and tablets). Nothing is uploaded.
The Control Panel shows the Source noise floor as soon as the file loads: the level of the quietest moments, where only background noise remains. A clean studio recording might read -70 dBFS; a laptop microphone in a noisy room might read -40.
Choose the settings
Start with the defaults, then adjust one setting at a time.
| Setting | Range (default) | Raise it when | Lower it when |
|---|---|---|---|
| Strength | 0 to 100% (70%) | Noise is still clearly audible | Speech sounds watery or thin |
| Noise floor | -80 to -20 dB (-45) | Noise survives in the pauses | Quiet parts of speech are damaged |
Remove, measure, compare
Select Remove noise. The Control Panel adds the Result noise floor and the Reduction in dB. A reduction of 10 dB or more is clearly audible; 15 to 20 dB is a big change.
Switch between Original and Result; the player stays at the same moment, so you hear the same words both ways. Listen to speech, not just the pauses: the best setting is the strongest one that does not change the voice.
Export
The output starts in your file's own format and quality. Change it if the file is headed somewhere specific, then select Download.
| Output | Best for | Note |
|---|---|---|
| WAV | Further editing, archiving | Lossless; largest files |
| FLAC | Archiving at a smaller size | Lossless; not every app plays it |
| MP3 | Sharing and uploading | Keep 128 kbps or higher for music |
| M4A / AAC | Apple devices, small files | Depends on browser support |
| OGG | Web and games | Depends on browser support |
Common jobs, step by step
A few jobs come up again and again.
- Laptop microphone hiss: defaults, then raise strength to 80% if hiss remains.
- Air conditioning or fan: noise floor -40 dB, strength 70%.
- Electrical hum: strength 80%; if it remains, raise the noise floor a little.
- Interview with two microphones: clean each recording before joining them with the Audio Joiner.
- Podcast workflow: Noise Remover, then Silence Remover, then Audio Normalizer to -16 LUFS.
What it does not do
This is spectral subtraction, a classic signal-processing method, not an AI voice model. It cannot separate a voice from music, crowd noise or another speaker, and very strong settings add a watery artifact.
For those cases a dedicated AI tool will do better. For steady background noise, which is the most common problem, this approach is fast, private and predictable.
Written by
Ankit Kumar
Senior Software Engineer & Founder at ToolBuddy
Engineering browser-first utilities with zero upload requirements using WebAssembly, Web Audio, and Canvas technologies.