@audio/denoise
audiojs.dev
- Free plan
- Open source
- Web
What @audio/denoise is
@audio/denoise is a free web tool for cleaning recordings and listening to or saving the result. Its browser-based processing uses classical signal processing rather than AI noise removal, and the page says audio stays on the device. Automatic mode measures the recording, selects a specialized method, and displays its plan; users can also choose methods such as dehum, declick, declip, and gate themselves. The tool describes thirteen open-source restoration methods and a classifier. It targets steady background noise, mains hum, clicks, pops, vinyl crackle, clipping, wind rumble, plosives, sibilance, room reverb, and noise between phrases. For music, it can address hum, clicks, crackle, and clipping; broadband denoising should be used lightly and compared with the original. Machine-learning models may do better with heavy speech-over-noise. Automatic mode does not select dereverb, which must be chosen explicitly. Developers can install the associated package with npm and process mono Float32Array audio.
Who it suits
It suits people who want to clean recordings in a browser without uploading audio, including users working on music or common recording noise. Developers can also use its npm package with mono Float32Array audio.
Good to know, and what to watch
- Runs in the browser; audio stays on the device.
- Automatic mode selects a method and shows its plan.
- Targets hum, clicks, crackle, clipping, and other noise.
- Manual method selection is available.
- Free plan; no account required.
- Automatic mode does not select dereverb.
- Machine-learning models may do better with heavy speech-over-noise.
- Broadband denoising should be used lightly and compared with the original.
- No batch processing.
Plans
Every plan and priceFree
Free browser noise remover
Free plan
Getting started
- Open the @audio/denoise website
- Load a recording in the browser
- Use automatic mode or choose a restoration method manually
- Listen to the processed result
- Save the result
Limits to check
There is no batch processing. Automatic mode does not choose dereverb, and heavy speech-over-noise may be handled better by machine-learning models.
The Club's review
Overview
@audio/denoise is a free browser tool for cleaning recordings and saving the repaired audio. It is best for people who want targeted fixes without sending audio off-device. Its specialized methods cover common recording flaws, though heavy speech-over-noise is a better reason to seek machine-learning cleanup.
Key features
Thirteen open-source restoration methods sit alongside a classifier. In automatic mode, the classifier measures the recording, chooses a method and displays its plan; that gives newcomers a suggested route while leaving the actual choice visible. Users who know the problem can instead select a method manually, including dehum, declick, declip and gate.
The target list ranges from steady background noise and mains hum to clicks, pops, vinyl crackle, clipping, wind rumble, plosives, sibilance, room reverb and noise between phrases. That breadth makes it useful for both speech recordings and music repair. The underlying processing is classical signal processing, not AI noise removal, so it is not the strongest choice when speech is buried under heavy noise; machine-learning models may do better there.
Automatic mode does not choose dereverb. Users who need room-reverb treatment must select it explicitly, which is a small but important distinction for anyone relying on the classifier to handle the entire job. For music, hum, clicks, crackle and clipping are sensible targeted repairs. Broadband denoising deserves more care: compare the result with the original and use it lightly.
All processing happens in the browser, and audio does not leave the device. That is a meaningful advantage for private recordings, as well as a simpler setup than sending files to a service. Developers can install the package with npm; its documented input and output are mono Float32Array audio, making it more suited to an audio-processing workflow than a general-purpose desktop editing environment.
Pricing
The Free plan costs nothing and runs in the browser with no account and no upload. It includes noise reduction, click and crackle removal, hum removal, declip repair and spectral repair. Batch processing is not included, so users handling collections of recordings should expect to work without a batch workflow. There are no paid tiers to weigh against that limitation.
Platforms
@audio/denoise is available on the web. Its browser-based processing avoids uploads; the package is also documented for npm installation and mono Float32Array processing, a separate route for developers integrating it into their own work.
Who it's for
This is a good fit for individuals who want to repair one recording at a time, especially where privacy, hum, clicks, crackle or clipping matter more than batch tools. Musicians can use its targeted repairs, with restraint around broadband denoising. It is less suited to users who need automated dereverb or advanced removal of speech masked by substantial noise.
Pros and cons
Pros
- On-device processing keeps audio from leaving the browser session.
- Automatic recommendations and manual method selection serve both beginners and users with a specific defect to fix.
- The free plan covers several practical repairs, including hum, clicks, crackle, clipping and spectral repair.
Cons
- No batch processing makes it less convenient for large sets of recordings.
- Automatic mode does not select dereverb, requiring a deliberate manual choice.
- Classical signal processing may be outperformed by machine-learning models on speech buried in heavy noise.
Alternatives
Audio Restoration Software is a useful directory for comparing a wider set of restoration options. For music-specific workflows, Music Transcription Software is a separate category to consider.
VinylRest is another free option for readers comparing restoration tools across Linux, macOS and web. Choose Vinyl Restoration Suite if you want a free restoration tool for Linux, macOS or Windows. Wave Corrector is a free alternative for Linux or Windows users.
Readers willing to pay once for plug-ins may prefer Wave Arts Master Restoration Suite 6, priced at $99, paid once and requiring a host that supports audio plug-ins. Accentize DeRoom is a paid alternative focused on reverb removal and room resonance suppression. Algorithmix Sound Rescue is another paid restoration option, with Sound Rescue 3.0 at $79.90, paid once.
Neiro is a free alternative available across Linux, macOS, self-hosted, web and Windows. For spectral audio editing with AI-assisted processing, consider SpectraLayers Pro.
Verdict
Choose @audio/denoise for private, no-cost cleanup of common recording problems when you value on-device processing and control over the repair method. Its range of focused fixes and automatic starting point are compelling for individual recordings; look elsewhere for batch processing or speech overwhelmed by noise.
The verdict
@audio/denoise offers a free, on-device way to address several kinds of recording noise, with automatic and manual method selection. Heavy speech-over-noise may call for machine-learning models, and dereverb requires manual selection.
Questions people ask
Is @audio/denoise free?
Yes. The Free plan costs nothing and runs in the browser without an account or upload.
Does the tool upload recordings?
No. The page says processing runs in the browser and nothing is uploaded or leaves the device.
Which noise problems can it address?
Listed targets include steady background noise, mains hum, clicks, pops, vinyl crackle, clipping, wind rumble, plosives, sibilance, room reverb, and noise between phrases.
Can I choose the restoration method?
Yes. Automatic mode selects a method from measured recording properties, or you can choose methods such as dehum, declick, declip, and gate.
Is there an npm package?
Yes. The related @audio/denoise package can be installed with npm and processes mono Float32Array audio in and out.
What the maker publishes
- Automatic API
- The denoise function can return cleaned samples, a classifier plan, or be forced to a named method. source
- Automatic classifier
- The classifier evaluates hum, click density, rumble, noise-floor stability and sibilance before selecting a method. source
- Automatic mode
- A classifier selects a specialized method based on measured properties of the recording and shows its plan. source
- Automatic mode limit
- The repository says automatic mode does not select dereverb; users must choose it explicitly. source
- Browser workflow
- Users can drop or paste a recording, inspect the selected method and plan, audition the result and save the cleaned file. source
- Classical DSP
- The product states that its methods use classical signal processing rather than AI models. source
- Click repair
- Clicks, pops and crackle are detected and reconstructed with LPC-based interpolation. source
- Clipping repair
- The package can detect clipping levels automatically and reconstruct clipped waveform sections. source
- Developer demo
- The repository includes a self-contained demo.html for selecting scenarios, choosing methods, viewing waveforms and hearing differences. source
- Developer use
- The repository documents installing the package with npm and processing mono Float32Array audio in and out. source
- Hiss processing
- Hiss and fan noise can be treated with Wiener or MMSE-LSA processing, or OM-LSA when speech is present. source
- Hum processing
- The dehum method tracks 50 or 60 Hz mains frequencies and their harmonics with adaptive notch or sinusoidal subtraction processing. source
- Input and output
- The JavaScript API accepts and returns mono Float32Array audio; stereo should be processed channel by channel. source
- Installation
- Installation is documented with npm using `npm install @audio/denoise`. source
- License
- The repository and npm metadata state that the package is released under the MIT license. source
- Limit
- The page says machine learning models can do better for heavy speech-over-noise. source
- Limit: automatic dereverb
- Automatic routing does not select dereverb because the documentation says it lacks a reliable single-pass signature; dereverb must be selected explicitly. source
- Limit: data shape
- The core API is sample-array based and does not provide a hosted upload service or graphical desktop editor in the package documentation. source
- Limit: heavy speech noise
- The maker cautions that machine-learning models can outperform these classical methods for heavy speech-over-noise cases. source
- Limit: single-pass operations
- The package describes itself as single-pass noise reduction, and several operations such as desqueak are documented as whole-clip rather than streaming processors. source
- Maker history
- The home page dates audiojs maintenance to 2016 and describes the stack as cross-runtime JavaScript for browser, Node, Deno and Bun. source
- Maker scale
- The audiojs home page states more than 300 open-source packages and over 9 million npm downloads per month. source
- Manual selection
- Users can choose methods such as dehum, declick, declip and gate by hand. source
- Methods
- The page describes thirteen open-source restoration methods plus a classifier. source
Checked against the maker's pages on 15 Oct 2026.