Skip to main content
v1.1.1 โ€” iOS

Enterprise NoiseCancellation for

NCKit-powered audio intelligence for iOS & Android. Real-time & offline. 100% on-device. Zero cloud dependency.

iOS 16+Swift 5.9+arm64On-DeviceSPM ยท CocoaPods
DenoiseService.swift
import NCKit

// 1. Load bundled NCKit model
let modelURL = try NCKitModelLocator.modelTarGzURL()

// 2. Create processor โ€” reuse across files
let processor = try NCKitProcessor(
    modelURL: modelURL,
    attenLimDb: 100
)

// 3. Denoise โ€” streaming I/O, any file length
try NCKitFileProcessor.processFile(
    inputURL:  noisyRecording,
    outputURL: cleanOutput,
    processor: processor
)
0x
Faster than realtime
0ms
Processing latency
0%
On-device, no cloud
0MB
Model size

Everything you need for
professional audio

A complete on-device audio intelligence stack โ€” from neural inference to output normalization.

๐Ÿง 

NCKit Engine

State-of-the-art GRU neural network with ERB filterbank processing. Same intelligence used in professional studio tools โ€” now on your iPhone.

๐Ÿ“ฑ

100% On-Device

Audio never leaves the device. GDPR & HIPAA-friendly by architecture. Works fully offline.

โšก

Sub-20ms Latency

Real-time frame processing with pre-allocated buffers. Zero glitches.

๐Ÿ—‚

Offline File Processing

Denoise any AVFoundation-readable file. Streaming I/O โ€” safe for recordings of any length on-device.

๐Ÿ”Š

Smart Normalization

Speech-gated makeup gain + tanh soft limiter. No pumping, consistent output.

๐Ÿ”’

Type-Safe & Sendable

NCKitError is structured and Sendable โ€” works seamlessly in async/await and across actor boundaries.

๐Ÿ“ฆ

SPM ยท CocoaPods ยท XCFramework

All three distribution methods supported out of the box.

๐Ÿ”ง

Runtime Configurable

Adjust attenLimDb and postFilterBeta at runtime โ€” no model reload required.

Real-world audio
environments

๐Ÿ—

Field Inspections

HVAC, outdoor, construction sites

๐Ÿ“ž

VoIP & Conferencing

Crystal-clear calls from anywhere

๐ŸŽ™

Podcast & Recording

Studio quality on any microphone

๐Ÿฅ

Telehealth

Intelligible audio for critical care

๐Ÿš—

In-Vehicle

Road noise isolation

๐ŸŽ“

EdTech

Clear voice in noisy classrooms

Hear the
difference

Compare noisy input against NCKit-processed output. For interactive A/B testing, try the Sample App.

Noisy input

Original recording with background noise

NCKit output

After NCKit denoising

The audio pipeline,
simplified

๐ŸŽค
Input
Any AVFoundation format
โ†’
๐Ÿ”„
Resample
โ†’ 48 kHz mono Float32
โ†’
๐Ÿง 
NCKit
GRU neural inference
โ†’
โœจ
Normalize
Speech-gated gain
โ†’
๐Ÿ“„
Output
16-bit WAV, 48 kHz

Three ways to
integrate

Pick the method that fits your workflow. All three are fully supported.

Full Swift Package Manager Guide โ†’
swift
.package(
  url: "https://github.com/
       5Exceptions-Mobile-Team/NCKit.git",
  exact: "1.1.1"
)

Ready to ship
studio-quality audio?

Get started in under 5 minutes. Integrate NCKit and transform your app's audio experience.

On-device processing ยท No cloud upload ยท iOS 16+