Implement silero-based audio chunker(#857) & introducing adaptive VAD & fix compilations on Windows(#66)/Linux(#67)

.aiignore

@@ -0,0 +1,7 @@
+# An .aiignore file follows the same syntax as a .gitignore file.
+# .gitignore documentation: https://git-scm.com/docs/gitignore
+# Junie will ask for explicit approval before view or edit the file or file within a directory listed in .aiignore.
+# Only files contents is protected, Junie is still allowed to view file names even if they are listed in .aiignore.
+# Be aware that the files you included in .aiignore can still be accessed by Junie in two cases:
+# - If Brave Mode is turned on.
+# - If a command has been added to the Allowlist — Junie will not ask for confirmation, even if it accesses - files and folders listed in .aiignore.

.coderabbit.yaml

@@ -6,10 +6,53 @@ reviews:
   high_level_summary: false
   collapse_walkthrough: true
   path_instructions:
-    - path: "**/*.{js,ts,tsx,rs}"
+    - path: "**/*.{js,ts,tsx}"
       instructions: |
-        1. No error handling.
-        2. No unused imports, variables, or functions.
-        3. For comments, keep it minimal. It should be about "Why", not "What".
+        TypeScript/React Review Guidelines:
+        1. Ensure proper error handling with try-catch blocks or error boundaries
+        2. Check for unused imports, variables, or functions
+        3. Verify TypeScript strict mode compliance
+        4. Validate proper use of React hooks (dependencies, cleanup)
+        5. Check for performance issues (unnecessary re-renders, missing memoization)
+        6. Ensure consistent use of functional components
+        7. Verify proper async/await usage and promise handling
+        8. Check for accessibility concerns in UI components
+        9. Comments should explain "why" not "what" - remove obvious comments
+        10. Ensure consistent naming: kebab-case files, PascalCase components
+
+    - path: "**/*.rs"
+      instructions: |
+        Rust Review Guidelines:
+        1. Ensure proper error handling with Result/Option types
+        2. Check for clippy warnings and suggest fixes
+        3. Verify memory safety and absence of unnecessary unsafe blocks
+        4. Check for proper use of lifetimes and borrowing
+        5. Ensure efficient use of iterators over manual loops
+        6. Verify proper async/await usage with Tokio
+        7. Check for appropriate use of Arc/Mutex in concurrent code
+        8. Ensure consistent error types using thiserror
+        9. Verify proper use of tracing for logging
+        10. Comments should explain "why" not "what"
+        11. Check adherence to Rust Style Guide (enforced by rustfmt)
+        12. Verify performance-critical paths use appropriate optimizations
+
+    - path: "**/crates/audio/**/*.rs"
+      instructions: |
+        Audio Processing Specific:
+        1. Verify real-time constraints are met (no blocking operations)
+        2. Check for proper buffer management and zero-copy where possible
+        3. Ensure platform-specific code is properly feature-gated
+        4. Verify sample rate conversions are handled correctly
+        5. Check for potential audio artifacts or discontinuities
+
+    - path: "**/plugins/**/*.rs"
+      instructions: |
+        Tauri Plugin Specific:
+        1. Ensure commands are properly exposed with #[tauri::command]
+        2. Verify error types are serializable for IPC
+        3. Check for proper permission handling
+        4. Ensure TypeScript bindings will generate correctly
+        5. Verify async commands use proper runtime handling
+
 chat:
   auto_reply: false

.cursor/rules/audio-processing.mdc

@@ -0,0 +1,70 @@
+---
+description: Audio processing pipeline patterns and best practices
+globs:
+  - "crates/audio/**/*.rs"
+  - "crates/chunker/**/*.rs"
+  - "crates/vad/**/*.rs"
+  - "crates/aec/**/*.rs"
+  - "crates/denoise/**/*.rs"
+  - "crates/stt*/**/*.rs"
+alwaysApply: false
+---
+
+# Audio Processing Pipeline
+
+## Architecture Overview
+Real-time audio capture → VAD → Echo cancellation → Chunking → STT
+
+## Key Components
+
+### Audio Capture (`crates/audio/`)
+- Platform-specific implementations:
+  - macOS: CoreAudio
+  - Windows: WASAPI
+  - Linux: ALSA
+- Zero-copy operations for performance
+- Stream-based processing
+
+### Voice Activity Detection (`crates/vad/`)
+- Silero VAD with ONNX runtime
+- Minimum 480 samples (30ms at 16kHz) requirement
+- Confidence thresholds based on speech physics
+
+### Audio Chunking (`crates/chunker/`)
+- SmartPredictor for advanced feature analysis
+- Multi-feature fusion: VAD + Speech Quality + SNR
+- Fallback chain: SmartPredictor → Silero → RMS
+
+### Echo Cancellation (`crates/aec/`, `crates/aec2/`)
+- Multiple AEC implementations
+- Real-time processing requirements
+
+### Speech-to-Text (`crates/stt*`)
+- Unified interface in `crates/stt/`
+- Multiple backends (Local-first for privacy):
+  - Local: Whisper (with Metal/CUDA acceleration)
+- Stream-based transcription
+
+## Performance Guidelines
+- Use stream processing for real-time data
+- Avoid memory allocations in hot paths
+- Platform-specific SIMD optimizations
+- ONNX GraphOptimizationLevel::Level3
+- Chunk-based processing for long sessions
+
+## Error Handling Patterns
+```rust
+#[derive(thiserror::Error, Debug)]
+pub enum AudioError {
+    #[error("Buffer underrun")]
+    BufferUnderrun,
+
+    #[error("Device not available: {0}")]
+    DeviceError(String),
+}
+```
+
+## Testing Considerations
+- Use `serial_test` for audio device tests
+- Mock audio streams for unit tests
+- Integration tests with sample audio files