[02/09] audio: capture lifecycle fix + ALSA stderr suppression

config/README.md

@@ -0,0 +1,120 @@
+# Application Configuration
+
+This directory contains the configuration files for the ColdVox application.
+
+## `default.toml`
+
+This is the primary configuration file for the application. It contains the default settings for all components, including text injection, VAD, and STT.
+
+The application loads this file at startup. The values in this file can be overridden by environment variables or command-line arguments.
+
+## Security Best Practices
+
+**Important Security Note:** Do not store secrets, API keys, passwords, or any sensitive information in `default.toml` or any committed configuration files. This file is intended for default, non-sensitive values only and should be version-controlled.
+
+- Use environment variables for overriding sensitive values (e.g., `COLDVOX_STT__PREFERRED=your_secret_plugin`). Refer to [docs/user/runflags.md](docs/user/runflags.md) for all overridable variables.
+- For local development or production overrides, create a `config/overrides.toml` file with your custom settings. Add `config/overrides.toml` to your `.gitignore` to prevent accidental commits of sensitive data.
+- If implementing custom loading, you can extend the config builder in `crates/app/src/main.rs` to include `overrides.toml` after `default.toml` for layered overrides.
+- Always validate and sanitize configuration values at runtime to prevent injection attacks or invalid settings.
+
+Example `overrides.toml` template (create this file for local use):
+
+```toml
+# Local overrides for default.toml - add to .gitignore!
+# This file is not loaded by default; extend Settings::new() if needed.
+
+# Example: Override injection settings
+[Injection]
+fail_fast = true  # Maps to COLDVOX_INJECTION__FAIL_FAST=true
+max_total_latency_ms = 500  # Maps to COLDVOX_INJECTION__MAX_TOTAL_LATENCY_MS=500
+
+# Example: STT preferences (avoid committing model paths with secrets)
+[stt]
+preferred = "local_whisper"  # Maps to COLDVOX_STT__PREFERRED=local_whisper
+max_mem_mb = 2048  # Maps to COLDVOX_STT__MAX_MEM_MB=2048
+```
+
+## Deployment Considerations
+
+When deploying ColdVox, handle configurations carefully to ensure security, flexibility, and reliability across environments.
+
+### Including config/default.toml in Builds and Deployments
+- **Repository**: Always commit `config/default.toml` as it holds safe, default values. Do not modify it for environment-specific needs.
+- **Build Process**: The TOML is loaded at runtime, not embedded. In CI/CD (e.g., via `cargo build --release`), copy `config/default.toml` to the deployment artifact or container.
+  - Example in Dockerfile:
+    ```
+    COPY config/default.toml /app/config/
+    COPY target/release/coldvox-app /app/
+    WORKDIR /app
+    CMD ["./coldvox-app"]
+    ```
+  - For binary distributions: Include in a `config/` subdirectory next to the executable.
+- **Runtime Loading**: The app loads `config/default.toml` relative to the working directory. XDG support not implemented; to add it, extend `Settings::new()` with XDG path lookup (see deployment docs for details).
+
+### Environment-Specific Configurations
+- **Overrides via Environment Variables**: Preferred for secrets and dynamic settings. Use `COLDVOX__` prefix:
+  - Example for production: `export COLDVOX_STT__PREFERRED=cloud_whisper; export COLDVOX_INJECTION__FAIL_FAST=true`.
+  - Nested: `COLDVOX_VAD__SENSITIVITY=0.8` overrides `[vad].sensitivity`.
+  - Set in deployment tools: Systemd (`Environment=`), Docker (`-e`), Kubernetes (Secrets/ConfigMaps).
+- **Separate TOML Files for Non-Secrets**: Use `overrides.toml` (or env-specific like `staging.toml`) for bulk overrides. Extend the loader in `crates/app/src/main.rs` to support `COLDVOX_CONFIG_OVERRIDE_PATH=/path/to/staging.toml`.
+  - Template extension for staging:
+    ```toml
+    # staging.toml - non-sensitive overrides
+    [stt]
+    preferred = "vosk"
+    language = "en"
+
+    [injection]
+    injection_mode = "keystroke"  # Staging: Test keystroke reliability
+    ```
+  - Current load order: CLI flags > Env vars > default.toml > hardcoded defaults. Note: `overrides.toml` is a template and NOT automatically loaded. To enable, add `.add_source(File::with_name("config/overrides.toml").required(false))` to `Settings::new()`.
+- **Validation**: On deploy, validate configs (see [docs/deployment.md](docs/deployment.md) for steps, including parsing checks and tests).
+
+### Best Practices
+- **Secrets Management**: Use tools like HashiCorp Vault, AWS Secrets Manager, or env files (`.env` with `dotenv` if extended).
+- **Rollback**: Backup configs before deploy; fallback to env vars if TOML fails.
+- **CI Integration**: Test config loading in workflows (e.g., set mock env vars in `.github/workflows/ci.yml`).
+- For full deployment details, including validation and rollback, refer to [docs/deployment.md](docs/deployment.md).
+
+## `plugins.json`
+
+This file contains the configuration for the STT (Speech-to-Text) plugin manager. It defines the preferred plugin, fallback plugins, and other settings related to plugin management.
+
+While the main application configuration is in `default.toml`, this file is kept separate to potentially allow for dynamic updates or for management by external tools in the future.
+
+## For Test Authors
+
+Tests that need to load configuration should use `Settings::from_path()` with `CARGO_MANIFEST_DIR`:
+
+```rust
+#[cfg(test)]
+use std::env;
+use std::path::PathBuf;
+
+fn get_test_config_path() -> PathBuf {
+    // Try workspace root first (for integration tests)
+    let workspace_config = PathBuf::from(env!("CARGO_MANIFEST_DIR"))
+        .parent()
+        .unwrap()
+        .parent()
+        .unwrap()
+        .join("config/default.toml");
+
+    if workspace_config.exists() {
+        return workspace_config;
+    }
+
+    // Fallback to relative path from crate root
+    PathBuf::from(env!("CARGO_MANIFEST_DIR"))
+        .join("../../config/default.toml")
+}
+
+#[test]
+fn my_test() {
+    let config_path = get_test_config_path();
+    let settings = Settings::from_path(&config_path)?;
+    // ... test logic
+}
+```
+
+This ensures tests work regardless of working directory context.

config/default.toml

@@ -0,0 +1,61 @@
+# ColdVox default configuration file
+# Root-level app settings
+resampler_quality = "balanced"       # "fast", "balanced", "quality"
+activation_mode = "vad"              # "vad", "hotkey"
+enable_device_monitor = true
+# device =  "Device Name"            # Optional: specific device (omit for default)
+
+[injection]
+# Core behavior
+fail_fast = false                # Exit immediately if all injection methods fail
+allow_kdotool = false            # Enable kdotool fallback (KDE/X11)
+allow_enigo = false              # Enable enigo fallback (input simulation)
+inject_on_unknown_focus = true   # Allow injection when focus is unknown
+require_focus = false            # Require editable focus for injection
+pause_hotkey = ""                # Hotkey to pause/resume injection (e.g., "Ctrl+Alt+P")
+redact_logs = true               # Redact text in logs for privacy
+
+# Timing and latency
+max_total_latency_ms = 800       # Max latency for a single injection call (ms)
+per_method_timeout_ms = 250      # Timeout for each method attempt (ms)
+paste_action_timeout_ms = 200    # Timeout for paste actions (ms)
+
+# Cooldown/backoff
+cooldown_initial_ms = 10000      # Initial cooldown after failure (ms)
+cooldown_backoff_factor = 2.0    # Exponential backoff factor
+cooldown_max_ms = 300000         # Max cooldown period (ms)
+
+# Injection behavior
+injection_mode = "auto"          # "keystroke", "paste", or "auto"
+keystroke_rate_cps = 20          # Keystroke rate (chars/sec)
+max_burst_chars = 50             # Max chars per burst
+paste_chunk_chars = 500          # Chunk size for paste ops
+chunk_delay_ms = 30              # Delay between paste chunks (ms)
+
+# Focus/window management
+focus_cache_duration_ms = 200    # Cache duration for focus status (ms)
+enable_window_detection = true   # Enable window manager integration
+clipboard_restore_delay_ms = 500 # Delay before restoring clipboard (ms)
+discovery_timeout_ms = 1000      # Timeout for window discovery (ms)
+
+# App allow/block lists
+allowlist = []                   # List of allowed app patterns (regex)
+blocklist = []                   # List of blocked app patterns (regex)
+
+# Success rate tuning
+min_success_rate = 0.3           # Minimum success rate before fallback
+min_sample_size = 5              # Samples before trusting success rate
+
+[stt]
+# preferred = "vosk"              # Preferred STT engine (omit for auto-select)
+fallbacks = []
+require_local = false
+# max_mem_mb = 1024               # Memory limit in MB (omit for no limit)
+# language = "en-US"              # Language code (omit for default)
+failover_threshold = 5
+failover_cooldown_secs = 10
+model_ttl_secs = 300
+disable_gc = false
+metrics_log_interval_secs = 30
+debug_dump_events = false
+auto_extract = true

config/overrides.toml

@@ -0,0 +1,54 @@
+# Local overrides for default.toml - add config/overrides.toml to .gitignore!
+# This file is not loaded by default; extend Settings::new() in crates/app/src/main.rs to include it if desired.
+# Use this for local development or production settings without committing sensitive data.
+# All values here override those in default.toml and can also be overridden by env vars.
+
+# Example: Override general settings
+device = "your_preferred_mic"  # Maps to COLDVOX_DEVICE=your_preferred_mic
+resampler_quality = "quality"  # Maps to COLDVOX_RESAMPLER_QUALITY=quality
+enable_device_monitor = true   # Maps to COLDVOX_ENABLE_DEVICE_MONITOR=true
+activation_mode = "hotkey"     # Maps to COLDVOX_ACTIVATION_MODE=hotkey
+
+# Example: Override injection settings
+[Injection]
+fail_fast = true                       # Maps to COLDVOX_INJECTION__FAIL_FAST=true
+allow_kdotool = true                   # Maps to COLDVOX_INJECTION__ALLOW_KDOTOOL=true
+allow_enigo = false                    # Maps to COLDVOX_INJECTION__ALLOW_ENIGO=false
+inject_on_unknown_focus = false        # Maps to COLDVOX_INJECTION__INJECT_ON_UNKNOWN_FOCUS=false
+require_focus = true                   # Maps to COLDVOX_INJECTION__REQUIRE_FOCUS=true
+pause_hotkey = "Ctrl+Alt+P"            # Maps to COLDVOX_INJECTION__PAUSE_HOTKEY=Ctrl+Alt+P
+redact_logs = true                     # Maps to COLDVOX_INJECTION__REDACT_LOGS=true
+max_total_latency_ms = 500             # Maps to COLDVOX_INJECTION__MAX_TOTAL_LATENCY_MS=500
+per_method_timeout_ms = 200            # Maps to COLDVOX_INJECTION__PER_METHOD_TIMEOUT_MS=200
+paste_action_timeout_ms = 150          # Maps to COLDVOX_INJECTION__PASTE_ACTION_TIMEOUT_MS=150
+cooldown_initial_ms = 5000             # Maps to COLDVOX_INJECTION__COOLDOWN_INITIAL_MS=5000
+cooldown_backoff_factor = 1.5          # Maps to COLDVOX_INJECTION__COOLDOWN_BACKOFF_FACTOR=1.5
+cooldown_max_ms = 60000                # Maps to COLDVOX_INJECTION__COOLDOWN_MAX_MS=60000
+injection_mode = "paste"               # Maps to COLDVOX_INJECTION__INJECTION_MODE=paste
+keystroke_rate_cps = 15                # Maps to COLDVOX_INJECTION__KEYSTROKE_RATE_CPS=15
+max_burst_chars = 30                   # Maps to COLDVOX_INJECTION__MAX_BURST_CHARS=30
+paste_chunk_chars = 300                # Maps to COLDVOX_INJECTION__PASTE_CHUNK_CHARS=300
+chunk_delay_ms = 50                    # Maps to COLDVOX_INJECTION__CHUNK_DELAY_MS=50
+focus_cache_duration_ms = 100          # Maps to COLDVOX_INJECTION__FOCUS_CACHE_DURATION_MS=100
+enable_window_detection = false        # Maps to COLDVOX_INJECTION__ENABLE_WINDOW_DETECTION=false
+clipboard_restore_delay_ms = 300       # Maps to COLDVOX_INJECTION__CLIPBOARD_RESTORE_DELAY_MS=300
+discovery_timeout_ms = 800             # Maps to COLDVOX_INJECTION__DISCOVERY_TIMEOUT_MS=800
+allowlist = ["firefox", "chrome"]      # Maps to COLDVOX_INJECTION__ALLOWLIST=firefox,chrome
+blocklist = ["password_manager"]       # Maps to COLDVOX_INJECTION__BLOCKLIST=password_manager
+min_success_rate = 0.5                 # Maps to COLDVOX_INJECTION__MIN_SUCCESS_RATE=0.5
+min_sample_size = 10                   # Maps to COLDVOX_INJECTION__MIN_SAMPLE_SIZE=10
+
+# Example: STT preferences (avoid committing sensitive model paths or API keys)
+[stt]
+preferred = "vosk"                     # Maps to COLDVOX_STT__PREFERRED=vosk
+fallbacks = ["whisper", "mock"]        # Maps to COLDVOX_STT__FALLBACKS=whisper,mock
+require_local = true                   # Maps to COLDVOX_STT__REQUIRE_LOCAL=true
+max_mem_mb = 1024                      # Maps to COLDVOX_STT__MAX_MEM_MB=1024
+language = "en"                        # Maps to COLDVOX_STT__LANGUAGE=en
+failover_threshold = 5                 # Maps to COLDVOX_STT__FAILOVER_THRESHOLD=5
+failover_cooldown_secs = 60            # Maps to COLDVOX_STT__FAILOVER_COOLDOWN_SECS=60
+model_ttl_secs = 600                   # Maps to COLDVOX_STT__MODEL_TTL_SECS=600
+disable_gc = false                     # Maps to COLDVOX_STT__DISABLE_GC=false
+metrics_log_interval_secs = 30         # Maps to COLDVOX_STT__METRICS_LOG_INTERVAL_SECS=30
+debug_dump_events = true               # Maps to COLDVOX_STT__DEBUG_DUMP_EVENTS=true
+auto_extract = true                    # Maps to COLDVOX_STT__AUTO_EXTRACT=true

config/plugins.json

@@ -0,0 +1,20 @@
+{
+  "preferred_plugin": null,
+  "fallback_plugins": [],
+  "require_local": false,
+  "max_memory_mb": null,
+  "required_language": null,
+  "failover": {
+    "failover_threshold": 5,
+    "failover_cooldown_secs": 10
+  },
+  "gc_policy": {
+    "model_ttl_secs": 300,
+    "enabled": true
+  },
+  "metrics": {
+    "log_interval_secs": 30,
+    "debug_dump_events": false
+  },
+  "auto_extract_model": true
+}

crates/app/Cargo.toml

@@ -2,7 +2,10 @@
 name = "coldvox-app"
 version = "0.1.0"
 edition = "2021"
-default-run = "coldvox"
+
+[lib]
+name = "coldvox_app"
+path = "src/lib.rs"
 
 [[bin]]
 name = "coldvox"
@@ -88,6 +91,7 @@ coldvox-vad-silero = { path = "../coldvox-vad-silero", features = ["silero"] }
 coldvox-stt = { path = "../coldvox-stt", features = ["parakeet", "whisper"] }
 csv = "1.3"
 cpal = "0.16.0"
+config = { version = "0.14", features = ["toml"] }
 coldvox-stt-vosk = { path = "../coldvox-stt-vosk", optional = true }
 
 [dev-dependencies]

crates/app/src/audio/mod.rs

@@ -1,14 +1,3 @@
 pub mod vad_adapter;
 pub mod vad_processor;
-
-// Re-export modules from coldvox-audio crate
-pub use coldvox_audio::{
-    capture::CaptureStats,
-    chunker::{AudioChunker, ChunkerConfig, ResamplerQuality},
-    frame_reader::FrameReader,
-    ring_buffer::{AudioProducer, AudioRingBuffer},
-};
-
-pub use coldvox_audio::AudioFrame;
-pub use vad_adapter::*;
-pub use vad_processor::*;
+pub mod wav_file_loader;

crates/app/src/audio/vad_processor.rs

@@ -86,7 +86,7 @@ impl VadProcessor {
                         timestamp_ms,
                         energy_db,
                     } => {
-                        info!(
+                        debug!(
                             "VAD: Speech started at {}ms (energy: {:.2} dB)",
                             timestamp_ms, energy_db
                         );
@@ -96,7 +96,7 @@ impl VadProcessor {
                         duration_ms,
                         energy_db,
                     } => {
-                        info!(
+                        debug!(
                             "VAD: Speech ended at {}ms (duration: {}ms, energy: {:.2} dB)",
                             timestamp_ms, duration_ms, energy_db
                         );
@@ -126,14 +126,14 @@ impl VadProcessor {
 
         self.frames_processed += 1;
 
-        if self.frames_processed % 100 == 0 {
+        if self.frames_processed.is_multiple_of(100) {
             tracing::debug!(
                 "VAD: Received {} frames, processing active",
                 self.frames_processed
             );
         }
 
-        if self.frames_processed % 1000 == 0 {
+        if self.frames_processed.is_multiple_of(1000) {
             debug!(
                 "VAD processor: {} frames processed, {} events generated, current state: {:?}",
                 self.frames_processed,