This Android library provides a reusable, voice-driven AI conversation pipeline, enabling developers to quickly integrate voice assistant capabilities into their applications. Targeting Android developers, it simplifies the integration of speech-to-text, AI understanding via Anthropic Claude, and text-to-speech, all within an MVVM architecture and featuring Jetpack Compose UI components.

How It Works

The SDK orchestrates a multi-stage pipeline: capturing microphone audio, performing Voice Activity Detection (VAD), transcribing speech to text (STT) using pluggable engines, sending transcripts to Anthropic Claude for AI-driven responses, and converting replies back to speech (TTS). This process is managed by a VoiceAISession and configured via a builder pattern, offering flexibility with swappable STT/TTS engines and on-device emotion detection.

Quick Start & Requirements

• Installation: Add com.sdk:voice-ai-sdk:1.0.0 to build.gradle.kts.
• Permissions: Declare RECORD_AUDIO, INTERNET, ACCESS_NETWORK_STATE in AndroidManifest.xml.
• API Key: Store Anthropic API key in local.properties (e.g., ANTHROPIC_API_KEY=sk-ant-...) and expose via BuildConfig.
• Hilt: Annotate Application with @HiltAndroidApp and Activity with @AndroidEntryPoint if using Hilt.
• Prerequisites: Android Studio Meerkat+, Minimum SDK 24 (Android 7.0), Kotlin 2.0+ (project uses 2.3.21), Anthropic API key.

Highlighted Details

• Swappable Engines: Supports pluggable STT (Android built-in, Whisper) and TTS (Android built-in, ElevenLabs) engines.
• On-Device Emotion Detection: Analyzes audio for emotions and can inject this into the AI context for adaptive responses.
• Security Features: Includes PII redaction, encrypted key storage via Android Keystore, and optional certificate pinning.
• UI Components: Provides ready-to-use Jetpack Compose UI elements for voice interaction.

Maintenance & Community

No specific details on contributors, sponsorships, or community channels are provided in the README.

Licensing & Compatibility

• License: MIT License.
• Compatibility: Permissive MIT license allows for commercial use and integration into closed-source projects.

Limitations & Caveats

• External Dependencies: Core functionality relies on external APIs (Anthropic Claude, optionally Whisper, ElevenLabs), requiring API keys and network connectivity.
• Android Specific: Designed exclusively for the Android platform.