ferraus denoise

Remove steady background noise (hiss, hum, fan, room tone) or run a speech denoiser

Synopsis

ferraus denoise <input> [options] [-o <output>]

Learns what the noise sounds like from a noise-only region (a saved profile, --noise-range, or an automatically found quiet second) and subtracts that fingerprint from the whole file with a soft spectral gate that avoids the chirping 'musical noise' of simple gates. --mode adaptive tracks a changing noise floor with no profile; --mode ml runs a speech model (RNNoise built in, DeepFilterNet 3 through its model slot); --mode hybrid runs the profile pass and then the model on what remains. With no options it finds the quietest second and learns the noise from it. --reduction 0dB returns the input exactly.

This command is also an fx stage and a pipeline op: ferraus fx in.wav -o out.wav --fx "denoise:key=value".

Examples

Find the quietest second, learn the noise from it, and remove it

ferraus denoise take.wav --auto -o clean.wav

Apply a saved room profile with stronger reduction

ferraus denoise take.wav --profile room.noiseprof --reduction 18dB -o clean.wav

One shot from a noise-only lead-in, and keep what was removed for listening

ferraus denoise interview.wav --noise-range 0s..1.5s --residual removed.wav -o clean.wav

Follow a changing background (traffic, wind) with no profile

ferraus denoise street.wav --adaptive --reduction 10dB -o clean.wav

Speech denoising with DeepFilterNet 3 (downloads the 8 MB model once with -y)

ferraus denoise call.wav --mode ml -y -o clean.wav

Speech denoising with the built-in RNNoise model (no download)

ferraus denoise call.wav --mode ml --model rnnoise -o clean.wav

Options

Flag	Type	Default	Meaning
`--profile`	path	—	Noise profile (.noiseprof JSON) learned with 'ferraus denoise profile'
`--noise-range`	range	—	One shot: learn the noise from this noise-only region of the input, e.g. 0s..2s
`--auto`	flag	—	One shot: find the quietest spectrally stable 1 s region and learn the noise from it (reported as noise_region in --json)
`--mode`	profile \| adaptive \| ml \| hybrid	`profile`	profile: spectral gate from a noise profile; adaptive: minimum-statistics tracking with no profile; ml: speech model; hybrid: profile pass, then the model
`--adaptive`	flag	—	Same as --mode adaptive
`--reduction`	gain	`12dB`	How far noise is pushed down, in dB (0dB leaves the input unchanged) Range: 0..60.
`--sensitivity`	number	`6`	Gate threshold above the noise profile: threshold = mean + sensitivity × std, scaled so 6 means 6 dB above steady hiss; higher removes more Range: 0..48.
`--smoothing`	integer	`3`	Smooth the gain mask across ± this many frequency bins (fewer chirps, slightly softer edges) Range: 0..32.
`--attack`	time	`20ms`	How fast the gate opens (a matching lookahead keeps onsets intact), e.g. 20ms
`--release`	time	`120ms`	How fast the gate closes after a sound ends, e.g. 120ms
`--artifact-control`	percent	`50%`	Blend in a median-filtered mask and a steady noise floor: higher removes more chirping at the cost of a little more residual noise
`--link`	on\|off	`on`	on: one gain mask for all channels so the stereo image stays put; off: per-channel masks
`--preserve-transients`	on\|off	`on`	Detect onsets and hold the gate open through them so attacks stay crisp
`--noisiness`	percent	`30%`	Adaptive mode: how much of the audio is noise; higher sets the threshold further above the tracked floor (20 dB × noisiness)
`--decay`	percent	`65%`	Adaptive mode: how slowly reduction relaxes after a sound (release = 200 ms × decay unless --release is given)
`--floor`	gain	`0dB`	Adaptive mode: shift the tracked noise floor in dB (positive removes more)
`--model`	dfn3 \| rnnoise \| music	`dfn3`	ml/hybrid modes: dfn3 = DeepFilterNet 3 (48 kHz, speech, best quality; model denoise/dfn3, 8 MB download), rnnoise = RNNoise (fast, light, built in, no download, speech), music = music-aware slot denoise/music-*
`--strength`	percent	`100%`	ml/hybrid modes: how much of the model's change to apply
`--residual`	path	—	Also write the removed component (input minus the fully processed output, before --mix) to this audio file
`--dereverb`	flag	—	Add the best-effort de-reverb stage (model slot denoise/dereverb-*)
`--mix`	percent	`100%`	Dry/wet balance: 0% returns the input unchanged, 100% is fully processed

Global options

Every command also accepts the global options (-o, --json, --dry-run, --overwrite, --progress, ...). See Global options.

Exit codes

Code	Meaning
0	Success
1	Processing failed
2	Usage or parameter error
3	Input missing or unreadable
4	Unsupported format or codec
5	Model not cached and downloads disabled
6	Model download or verification failed
8	Cancelled
10	Network error
70	Internal error (bug)

Common mistakes

Forgetting that outputs are never overwritten: pass --overwrite or a new -o path when re-running.
Beat-based times (4b, 1/4) need a tempo: pass --bpm or write 4b@120.
Percentages accept 80% or the fraction 0.8; a bare 80 means 80x, not 80%.
Gains need a unit when not in dB: 0.5x is a factor, -6 and -6dB are decibels.

For models

MCP tool: ferraus_denoise
Schema: ferraus schema denoise
Pipeline: { op: denoise, args: { ... } } (flag names with _ instead of -)
fx stage: yes

{"command": "denoise", "defaults": {"mode": "profile", "reduction": "12dB", "sensitivity": "6", "smoothing": "3", "attack": "20ms", "release": "120ms", "artifact_control": "50%", "link": "on", "preserve_transients": "on", "noisiness": "30%", "decay": "65%", "floor": "0dB", "model": "dfn3", "strength": "100%", "mix": "100%"}}