Enhance audio capture functionality and update dependencies

- Added support for capturing system audio on Windows using WASAPI with improved error handling and thread safety.
- Introduced the 'scopeguard' crate for better resource management during audio capture.
- Updated Cargo.toml to include 'scopeguard' and modified Windows-specific dependencies for enhanced functionality.
- Added a new test for validating audio capture output, ensuring the captured audio data is valid and non-empty.

backend/voicebox-server.spec

@@ -4,15 +4,16 @@ from PyInstaller.utils.hooks import collect_submodules
 from PyInstaller.utils.hooks import copy_metadata
 
 datas = []
-hiddenimports = ['backend', 'backend.main', 'backend.config', 'backend.database', 'backend.models', 'backend.profiles', 'backend.history', 'backend.tts', 'backend.transcribe', 'backend.utils.audio', 'backend.utils.cache', 'backend.utils.progress', 'backend.utils.hf_progress', 'backend.utils.validation', 'torch', 'transformers', 'fastapi', 'uvicorn', 'sqlalchemy', 'librosa', 'soundfile', 'qwen_tts', 'qwen_tts.inference', 'qwen_tts.inference.qwen3_tts_model', 'qwen_tts.inference.qwen3_tts_tokenizer', 'qwen_tts.core', 'qwen_tts.cli']
+hiddenimports = ['backend', 'backend.main', 'backend.config', 'backend.database', 'backend.models', 'backend.profiles', 'backend.history', 'backend.tts', 'backend.transcribe', 'backend.utils.audio', 'backend.utils.cache', 'backend.utils.progress', 'backend.utils.hf_progress', 'backend.utils.validation', 'torch', 'transformers', 'fastapi', 'uvicorn', 'sqlalchemy', 'librosa', 'soundfile', 'qwen_tts', 'qwen_tts.inference', 'qwen_tts.inference.qwen3_tts_model', 'qwen_tts.inference.qwen3_tts_tokenizer', 'qwen_tts.core', 'qwen_tts.cli', 'pkg_resources.extern']
 datas += collect_data_files('qwen_tts')
 datas += copy_metadata('qwen-tts')
 hiddenimports += collect_submodules('qwen_tts')
+hiddenimports += collect_submodules('jaraco')
 
 
 a = Analysis(
     ['server.py'],
-    pathex=['/Users/jamespine/Projects/voice/Qwen3-TTS'],
+    pathex=['C:\\Users\\ijame\\Projects\\voice\\Qwen3-TTS'],
     binaries=[],
     datas=datas,
     hiddenimports=hiddenimports,

tauri/src-tauri/Cargo.lock

@@ -4490,6 +4490,7 @@ dependencies = [
  "coreaudio-sys",
  "hound",
  "objc",
+ "scopeguard",
  "screencapturekit",
  "serde",
  "serde_json",

tauri/src-tauri/Cargo.toml

@@ -22,6 +22,7 @@ serde_json = "1.0"
 tokio = { version = "1", features = ["full"] }
 hound = "3.5"
 base64 = "0.22"
+scopeguard = "1.2.0"
 
 [target.'cfg(target_os = "macos")'.dependencies]
 screencapturekit = { version = "1", features = ["async"] }
@@ -31,7 +32,7 @@ core-foundation-sys = "0.8"
 
 [target.'cfg(target_os = "windows")'.dependencies]
 wasapi = "0.22"
-windows = { version = "0.62", features = ["Win32_Foundation", "Win32_UI_WindowsAndMessaging"] }
+windows = { version = "0.62", features = ["Win32_Foundation", "Win32_UI_WindowsAndMessaging", "Win32_System_Com"] }
 
 [target.'cfg(not(any(target_os = "android", target_os = "ios")))'.dependencies]
 tauri-plugin-updater = "2.0"

tauri/src-tauri/src/audio_capture/windows.rs

@@ -5,8 +5,8 @@ use std::io::Cursor;
 use std::sync::Arc;
 use std::sync::atomic::{AtomicBool, Ordering};
 use std::thread;
-use std::time::Duration;
 use wasapi::*;
+use windows::Win32::System::Com::{CoInitializeEx, CoUninitialize, COINIT_MULTITHREADED};
 
 pub async fn start_capture(
     state: &AudioCaptureState,
@@ -37,6 +37,20 @@ pub async fn start_capture(
     // Spawn capture task on a dedicated thread (WASAPI COM objects are not Send)
     // All WASAPI objects must be created and used on the same thread
     thread::spawn(move || {
+        // Initialize COM for this thread
+        unsafe {
+            let hr = CoInitializeEx(None, COINIT_MULTITHREADED);
+            if hr.is_err() {
+                eprintln!("Failed to initialize COM: {:?}", hr);
+                return;
+            }
+        }
+
+        // Ensure COM is uninitialized when thread exits
+        let _com_guard = scopeguard::guard((), |_| unsafe {
+            CoUninitialize();
+        });
+
         // Initialize WASAPI on this thread
         let device = match DeviceEnumerator::new()
             .and_then(|enumerator| enumerator.get_default_device(&Direction::Render))
@@ -76,10 +90,21 @@ pub async fn start_capture(
         *sample_rate_arc.lock().unwrap() = mix_format.get_samplespersec();
         *channels_arc.lock().unwrap() = mix_format.get_nchannels();
 
+        // Get device period
+        let (_def_period, min_period) = match audio_client.get_device_period() {
+            Ok(periods) => periods,
+            Err(e) => {
+                eprintln!("Failed to get device period: {}", e);
+                return;
+            }
+        };
+
         // Initialize audio client for loopback with StreamMode
+        // For loopback mode: get Render device, initialize with Capture direction
+        // This triggers AUDCLNT_STREAMFLAGS_LOOPBACK in the wasapi crate
         let stream_mode = StreamMode::EventsShared {
-            autoconvert: false,
-            buffer_duration_hns: 0, // 0 = use default buffer size
+            autoconvert: true,  // Enable automatic format conversion
+            buffer_duration_hns: min_period, // Use minimum period
         };
 
         if let Err(e) = audio_client.initialize_client(&mix_format, &Direction::Capture, &stream_mode) {
@@ -89,6 +114,15 @@ pub async fn start_capture(
             return;
         }
 
+        // Set up event handle for EventsShared mode
+        let h_event = match audio_client.set_get_eventhandle() {
+            Ok(event) => event,
+            Err(e) => {
+                eprintln!("Failed to set event handle: {}", e);
+                return;
+            }
+        };
+
         let capture_client = match audio_client.get_audiocaptureclient() {
             Ok(client) => client,
             Err(e) => {
@@ -158,8 +192,10 @@ pub async fn start_capture(
                 }
             }
 
-            // Sleep briefly to avoid busy-waiting
-            thread::sleep(Duration::from_millis(10));
+            // Wait for event signal (with timeout to allow checking stop flag)
+            if h_event.wait_for_event(100).is_err() {
+                // Timeout is expected - just continue to check stop flag
+            }
         }
 
         // Stop the stream when done

tauri/src-tauri/src/lib.rs

@@ -0,0 +1 @@
+pub mod audio_capture;

tauri/src-tauri/tests/audio_capture_test.rs

@@ -0,0 +1,59 @@
+// NOTE: This test requires system audio to be playing during execution.
+// To run this test successfully:
+//   1. Start playing audio (music, video, etc.)
+//   2. Run: cargo test --test audio_capture_test -- --nocapture
+//   3. The test will capture audio for 5 seconds and verify the output
+
+use voicebox::audio_capture::{AudioCaptureState, start_capture, stop_capture};
+use base64::Engine;
+
+#[tokio::test]
+async fn test_system_audio_capture() {
+    // Create AudioCaptureState
+    let state = AudioCaptureState::new();
+
+    println!("Starting system audio capture with 5 second max duration...");
+
+    // Start capture with 5 second max duration
+    let result = start_capture(&state, 5).await;
+
+    if let Err(e) = result {
+        panic!("Failed to start capture: {}", e);
+    }
+
+    println!("Capture started, waiting 5 seconds...");
+
+    // Wait 5 seconds for capture to complete
+    tokio::time::sleep(tokio::time::Duration::from_secs(5)).await;
+
+    println!("Stopping capture...");
+
+    // Stop capture and get the result
+    let audio_data = stop_capture(&state).await;
+
+    match audio_data {
+        Ok(base64_wav) => {
+            println!("Capture stopped successfully");
+
+            // Validate the returned base64 WAV data
+            println!("Validating base64 WAV data...");
+
+            // Decode base64 to bytes
+            let decoded_bytes = base64::engine::general_purpose::STANDARD
+                .decode(&base64_wav)
+                .expect("Failed to decode base64 data");
+
+            // Verify bytes array is not empty
+            assert!(!decoded_bytes.is_empty(), "Decoded bytes array is empty");
+
+            // Confirm data has content (length > 0)
+            println!("WAV data length: {} bytes", decoded_bytes.len());
+            assert!(decoded_bytes.len() > 0, "WAV data has no content");
+
+            println!("✓ Test passed: Audio capture produced valid WAV data");
+        }
+        Err(e) => {
+            panic!("Failed to stop capture or get audio data: {}", e);
+        }
+    }
+}