⚡ HUSHCORE | Local Real-Time Mic Processor

🌐 Türkçe Dokümantasyon için buraya tıklayın

HushCore is a local, privacy-first, low-latency audio processing application designed to enhance microphone inputs in real-time. It runs entirely on your local machine, capturing input, applying dynamic DSP filters (Gate, EQ, Compressor, Limiter) and Noise Suppression (Dual-Mode: Eco DSP or HQ AI), and routing the output to a virtual audio cable for consumption in Discord, OBS, Zoom, and other voice clients.

---

🚀 Features

• 100% Local & Private: No audio streams are sent to the cloud. Everything processed frame-by-frame on your CPU.
• Dual-Mode Noise Suppression:
  • Eco DSP: A lightweight, self-contained spectral subtraction algorithm with zero external AI dependencies. Runs at extremely low CPU footprint.
  • HQ AI: Incorporates DeepFilterNet deep learning models running on CPU for crystal-clear voice clarity.
• Studio FX Rack:
  • Noise Gate: Shuts down background static and key clicks when you are silent.
  • 3-Band Parametric EQ: Shape your voice warmth (Bass), clarity (Mids), and presence (Treble).
  • Dynamics Compressor: Evens out vocal dynamics so you never sound too quiet or clip.
  • Dynamic De-Esser: Attenuates harsh sibilant "S" and "T" sounds dynamically when sibilance spikes.
  • Harmonic Vocal Exciter: Infuses warm odd-harmonics (tanh drive) to add professional sparkling brightness ("air") and clarity.
  • Peak Limiter: Hard limiter protecting against digital clipping (>0 dBFS).
• Premium Feature Suite:
  • VST3 Host: Dynamically load any third-party .vst3 audio plugin. Opens the native plugin GUI editor directly on your desktop from the web panel.
  • Manual Latency Tuning: Set specific buffer sizes (64, 128, 256, 512 samples) to manually configure stream stability and latency.
  • 3D Cascade Visualizer: An offline, 60fps isometric waterfall spectrum display cascading backward in time with cyberpunk styling.
  • System Tray Launcher: Minimizes to the Windows system tray with options to toggle the audio engine, open the dashboard, or exit.
  • Dynamic API Key Security: Protects against unauthorized network access with startup token generation and request validation.
• Reflow-Free Level Meters: Highly optimized web interface using GPU accelerated CSS transformations to achieve smooth 60fps rendering without UI lag.
• Routing Status Alert Banner: Live UI warnings when outputting directly to physical speakers (feedback loops) vs routing successfully to Virtual Cable.

---

🛠️ Requirements

1. Node.js (v18 or newer) - Download
2. Python (3.8 - 3.12 recommended) - Download
3. VB-Audio Virtual Cable (Crucial to route audio to other apps) - Download

---

⚙️ Installation & Run

Windows (Quick Start)

Just double-click the run.bat file in the root folder. It will:

1. Install frontend dependencies and compile static production assets.
2. Create a Python Virtual Environment (venv) and install core dependencies.
3. Prompt you if you wish to download the PyTorch AI components (optional, ~1.5 GB).
4. Launch the local backend server.

---

🎧 Audio Routing Guide

To use HushCore as your microphone input in applications like Discord, Zoom, or OBS, you need to configure your audio routing correctly:

 [Physical Mic] ---> [HUSHCORE (Input)] 
                            |
                     (DSP & AI Processing)
                            |
                            v
                     [HUSHCORE (Output)] ---> [VB-Cable Input (Virtual)]
                                                        |
                                                 (Virtual Line)
                                                        |
                                                        v
                                              [VB-Cable Output] ---> [Discord/OBS/Zoom Input]

1. HushCore Dashboard Settings

1. Open http://127.0.0.1:8000 in your web browser.
2. Select your physical microphone in Microphone (Input Device).
3. Select CABLE Input (VB-Audio Virtual Cable) in Broadcast Output (Virtual Cable).
4. Click the ⚡ START ENGINE button.
5. Optional: Enable Hear Myself (Monitor) and select your physical headphones if you want to listen to your processed audio in real-time.

Warning

Why do I hear my own voice immediately? If you select physical speakers or headphones as the Broadcast Output, the processed audio will play back to you. Always route the output to VB-Cable Input and keep "Hear Myself" disabled unless you explicitly want to monitor your voice.

2. Client Application Configuration

In Discord, OBS, or Zoom:

1. Go to Settings -> Voice & Video.
2. Set your Input Device to CABLE Output (VB-Audio Virtual Cable).
3. Set your Output Device to your physical headphones/speakers.

---

💻 Tech Stack

• Backend: Python, FastAPI, WebSockets (Real-time meter & FFT streaming), Uvicorn.
• Audio DSP: NumPy, SciPy Signal, SoundDevice (PortAudio wrapper).
• HQ AI Engine: PyTorch, DeepFilterNet (Deep learning voice-enhancement).
• Frontend: Vanilla HTML5/CSS3 (Glassmorphic neon theme), Javascript (ES modules), Vite build compiler.

---

📄 License

Distributed under the MIT License. See LICENSE for more information.