Skip to content

MatissesProjects/EchoSense

Repository files navigation

EchoSense AI - Intelligent Adaptive Hearing Assistant

EchoSense is an advanced, low-latency Android application designed to act as a contextual hearing assistant. It leverages an on-device C++ DSP engine and Wear OS multi-mic array to isolate speech, enhance clarity, and provide real-time transcription.

Key Features

  • Low-Latency C++ Audio Engine: Powered by Google Oboe for high-performance, real-time audio processing.
  • Multi-Mic Array: Dynamically switch between Phone, Wired Headset, and Wear OS (Watch) microphones.
  • Intelligent DSP Pipeline:
    • Serial Routing: Optimized processing path (Voice Filters → Graphic EQ) to prevent gain spikes.
    • Soft-Knee Noise Gate: Includes a 50ms temporal hold (hysteresis) to prevent audio "chattering."
    • Remote AI Boost: Dedicated gain control for the Wear OS microphone to compensate for distance.
  • Audio Profiles: Instantly switch between Voice, Music, and TV optimized settings.
  • Live AI Transcription: On-device speech-to-text with persistent foreground processing.
  • Stealth Mode: Dimmed screen and "Keep Screen On" options for discreet, continuous use.

Tech Stack

  • Android NDK (C++17): Real-time DSP and filter implementation.
  • Kotlin: App logic, Foreground Services, and UI.
  • Wear OS Data Layer: High-speed PCM audio streaming from Watch to Phone.
  • Google Oboe: Low-latency AAudio/OpenSL ES wrapper.
  • Material Design 3: Modern, dark-themed interface for accessibility.

Build & Installation

Prerequisites

  • JDK 24+ (Required for the current build configuration).
  • Android SDK & NDK (v27+).
  • ADB enabled on both Phone and Watch.

Building from Source

Set your JAVA_HOME to a JDK 24+ installation:

$env:JAVA_HOME = 'C:\Path\To\JDK24'
./gradlew assembleDebug

Deploying to Devices

Connect your devices via ADB (Wireless Debugging recommended for Watch):

# Install to Phone
adb -s <phone_ip> install -r app/build/outputs/apk/debug/app-debug.apk

# Install to Watch
adb -s <watch_ip> install -r wear/build/outputs/apk/debug/wear-debug.apk

Roadmap

  • Phase 4: Room Database integration for conversation history and "Memory" profiles.
  • Phase 5 (Intelligent Scene Awareness):
    • Adaptive Scene Detection: Automatic switching based on acoustic fingerprinting.
    • Priority Profiles: Speech detection priority over ambient music.

Local

C:\Users\matis\AppData\Local\Android\Sdk\platform-tools\adb.exe connect 192.168.4.134:44759; C:\Users\matis\AppData\Local\Android\Sdk\platform-tools\adb.exe connect 192.168.4.159:41209; ./gradlew installDebug

About

Adaptive Hearing & Memory App

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors