aufklarer/DeepFilterNet3-CoreML

DeepFilterNet3 — Core ML (FP16)

Real-time speech enhancement model for Apple Silicon. Removes background noise from speech audio.

• 2.1M params, FP16, ~4.2 MB
• Runs on Neural Engine via Core ML
• 48kHz native, 10ms frames

Latency (M2 Max)

Duration	Time	RTF
5s	0.65s	0.13
10s	1.2s	0.12
20s	4.8s	0.24

Usage

import SpeechEnhancement

let enhancer = try await SpeechEnhancer.fromPretrained()
let clean = try enhancer.enhance(audio: noisyAudio, sampleRate: 48000)

swift run audio denoise noisy.wav --output clean.wav

Files

• DeepFilterNet3.mlpackage — Core ML FP16 model (Neural Engine)
• auxiliary.npz — ERB filterbank, Vorbis window, normalization states

Reference

• DeepFilterNet3
• Part of speech-swift

---

Links

• Blog: blog.ivan.digital
• Library Docs: soniqo.audio