### Install Rust and Tauri CLI

Source: https://github.com/aluzed/light-whisper/blob/main/README.md

Installs Rust using the official script and then installs the Tauri CLI for building Tauri applications.

```bash
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

cargo install tauri-cli --version "^2"
```

--------------------------------

### Install Linux Dependencies (Ubuntu)

Source: https://github.com/aluzed/light-whisper/blob/main/README.md

Installs necessary development libraries for building Tauri applications on Ubuntu 22.04+ or Debian 12+.

```bash
sudo apt-get install -y \
  libgtk-3-dev \
  libwebkit2gtk-4.1-dev \
  libayatana-appindicator3-dev \
  librsvg2-dev \
  libssl-dev \
  libasound2-dev \
  libxdo-dev \
  cmake \
  build-essential
```

--------------------------------

### Install Linux Dependencies (Fedora)

Source: https://github.com/aluzed/light-whisper/blob/main/README.md

Installs necessary development libraries for building Tauri applications on Fedora 39+.

```bash
sudo dnf install -y \
  gtk3-devel \
  webkit2gtk4.1-devel \
  libayatana-appindicator-gtk3 \
  librsvg2-devel \
  openssl-devel \
  alsa-lib-devel \
  libxdo-devel \
  cmake \
  gcc-c++
```

--------------------------------

### Install CMake on macOS

Source: https://github.com/aluzed/light-whisper/blob/main/README.md

Installs Homebrew if not already present, then installs CMake, a dependency for building C/C++ projects like whisper.cpp.

```bash
# Optional if you already have homebrew
/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

brew install cmake
```

--------------------------------

### GET get_config

Source: https://context7.com/aluzed/light-whisper/llms.txt

Retrieves the current application configuration including audio device, STT engine, model size, language, and keyboard shortcut.

```APIDOC
## GET get_config

### Description
Retrieves the current application configuration including audio device, STT engine selection, model size, language preference, and keyboard shortcut settings.

### Method
GET

### Endpoint
get_config

### Response
#### Success Response (200)
- **audio_device** (string) - Audio input device name or "default"
- **model_size** (string) - Whisper model: "tiny", "base", "small", "medium"
- **language** (string) - Language code: "auto", "en", "fr", etc.
- **engine** (string) - STT engine: "whisper" or "parakeet"
- **shortcut** (string) - Global hotkey in Tauri shortcut format

#### Response Example
{
  "audio_device": "default",
  "model_size": "base",
  "language": "auto",
  "engine": "whisper",
  "shortcut": "Alt+Space"
}
```

--------------------------------

### Get Application Configuration

Source: https://context7.com/aluzed/light-whisper/llms.txt

Retrieves the current application settings, including audio device, STT engine, model size, language, and hotkey. Use this to display current preferences or to initialize application state.

```javascript
const { invoke } = window.__TAURI__.core;

// Get current configuration
const config = await invoke('get_config');

// Returns: AppConfig object
// {
//   audio_device: "default",      // Audio input device name or "default"
//   model_size: "base",           // Whisper model: "tiny", "base", "small", "medium"
//   language: "auto",             // Language code: "auto", "en", "fr", etc.
//   engine: "whisper",            // STT engine: "whisper" or "parakeet"
//   shortcut: "Alt+Space"         // Global hotkey in Tauri shortcut format
// }

console.log(`Using engine: ${config.engine}`);
console.log(`Model size: ${config.model_size}`);
console.log(`Language: ${config.language}`);
```

--------------------------------

### GET list_audio_devices

Source: https://context7.com/aluzed/light-whisper/llms.txt

Returns a list of available audio input devices on the system.

```APIDOC
## GET list_audio_devices

### Description
Returns a list of available audio input devices on the system for microphone selection in settings.

### Method
GET

### Endpoint
list_audio_devices

### Response
#### Success Response (200)
- **devices** (array) - A list of strings representing available audio input device names.

#### Response Example
["MacBook Pro Microphone", "External USB Microphone", "AirPods Pro"]
```

--------------------------------

### Handle Recording Start/Stop Events

Source: https://context7.com/aluzed/light-whisper/llms.txt

Listens for 'recording-started' and 'recording-stopped' IPC events to update UI elements like timers and reset the waveform. A timer is started on 'recording-started' and cleared on 'recording-stopped'.

```javascript
const { event } = window.__TAURI__;

let timerInterval = null;
let startTime = null;

event.listen('recording-started', () => {
  startTime = Date.now();
  timerInterval = setInterval(() => {
    const elapsed = Math.floor((Date.now() - startTime) / 1000);
    const mins = Math.floor(elapsed / 60);
    const secs = elapsed % 60;
    document.getElementById('timer').textContent = `${mins}:${secs.toString().padStart(2, '0')}`;
  }, 200);
});

event.listen('recording-stopped', () => {
  clearInterval(timerInterval);
  document.getElementById('timer').textContent = '0:00';
});
```

--------------------------------

### Initialize and Use SttEngine

Source: https://context7.com/aluzed/light-whisper/llms.txt

Provides a unified interface for Whisper and Parakeet backends. Models must be loaded before transcription.

```rust
use crate::stt::SttEngine;
use std::path::Path;

// Create engine from config name
let mut engine = SttEngine::from_engine_name("whisper");  // or "parakeet"

// Or create specific engine type
let mut whisper = SttEngine::new_whisper();
let mut parakeet = SttEngine::new_parakeet();

// Load model
let model_path = Path::new("/Users/user/lightwhisper/models/ggml-base.bin");
engine.load_model(model_path)?;

// Check if model is loaded
if engine.is_loaded() {
    // Transcribe 16kHz mono f32 samples
    let samples_16k: Vec<f32> = /* resampled audio */;

    // Language: "auto", "en", "fr", etc. (Parakeet ignores this - auto-detects)
    let text = engine.transcribe(&samples_16k, "auto")?;
    println!("Transcription: {}", text);
}
```

--------------------------------

### Build and Run Commands

Source: https://github.com/aluzed/light-whisper/blob/main/CLAUDE.md

Standard commands for development, production, and debugging of the Tauri application.

```bash
# Development (hot reload frontend, debug backend)
cargo tauri dev

# Production build (creates .app + .dmg on macOS)
cargo tauri build

# Debug build (faster, no bundling optimization)
cargo tauri build --debug

# Build just the Rust backend (no frontend bundling)
cd src-tauri && cargo build
```

--------------------------------

### Build and Run Tauri Applications

Source: https://github.com/aluzed/light-whisper/blob/main/README.md

Commands for developing, building production, and debugging Tauri applications. Includes a command to build only the Rust backend.

```bash
# Development (hot reload frontend, debug backend)
cargo tauri dev

# Production build (creates .app + .dmg on macOS, .msi on Windows)
cargo tauri build

# Debug build (faster, no bundling optimization)
cargo tauri build --debug

# Rust backend only (no frontend bundling)
cd src-tauri && cargo build
```

--------------------------------

### Manage Application Configuration

Source: https://context7.com/aluzed/light-whisper/llms.txt

Handles loading and saving of JSON configuration files and provides helpers for directory paths.

```rust
use crate::config::{AppConfig, load_config, save_config_to_disk};

// Load config (returns defaults if file doesn't exist)
let config = load_config();

// Default values:
// AppConfig {
//     audio_device: "default",
//     model_size: "base",
//     language: "auto",
//     engine: "whisper",
//     shortcut: "Alt+Space",
// }

// Save modified config
let mut config = load_config();
config.engine = "parakeet".to_string();
config.language = "fr".to_string();
save_config_to_disk(&config)?;

// Directory paths
use crate::config::{config_dir, models_dir, parakeet_models_dir};

let cfg_dir = config_dir();        // ~/lightwhisper/
let models = models_dir();          // ~/lightwhisper/models/
let parakeet = parakeet_models_dir(); // ~/lightwhisper/models/parakeet-tdt/

// Ensure all directories exist
crate::config::ensure_dirs();
```

--------------------------------

### Manage Application Focus (macOS)

Source: https://context7.com/aluzed/light-whisper/llms.txt

Track and restore the previously focused application's PID before recording and activate it afterward. Ensure accessibility permissions are granted on macOS.

```rust
use crate::paste::{get_frontmost_pid, activate_pid};

// Get PID of currently focused application (before showing overlay)
let previous_pid = get_frontmost_pid();  // Returns i32, -1 if failed

// After recording/transcription, restore focus
if previous_pid > 0 {
    activate_pid(previous_pid);
}

// Check accessibility permission (macOS only)
use crate::paste::ensure_accessibility_permission;
let is_trusted = ensure_accessibility_permission();
// If false, macOS will prompt user to grant permission
```

--------------------------------

### Configuration Management

Source: https://context7.com/aluzed/light-whisper/llms.txt

Manages application configuration stored as JSON at ~/lightwhisper/config.json.

```APIDOC
## AppConfig

### Description
Handles loading and saving of application settings and directory management.

### Methods
- **load_config()**: Loads configuration from disk or returns defaults.
- **save_config_to_disk(config: &AppConfig)**: Persists modified configuration to disk.
- **config_dir()**: Returns the configuration directory path.
- **models_dir()**: Returns the models directory path.
- **ensure_dirs()**: Ensures all required application directories exist.
```

--------------------------------

### Storage Directory Structure

Source: https://github.com/aluzed/light-whisper/blob/main/CLAUDE.md

Layout of the application's configuration, model, and temporary file storage.

```text
~/lightwhisper/
├── config.json              # {audio_device, model_size, language}
├── models/ggml-{size}.bin   # Downloaded whisper.cpp models
└── temp/                    # Temporary WAV files (macOS/Linux)
```

--------------------------------

### Configure ParakeetEngine

Source: https://context7.com/aluzed/light-whisper/llms.txt

Interface for NVIDIA Parakeet TDT using ONNX Runtime. Requires a directory containing encoder, decoder, and vocabulary files.

```rust
use crate::stt::ParakeetEngine;
use std::path::Path;

let mut parakeet = ParakeetEngine::new();

// Load from directory containing encoder, decoder, and vocab files
let model_dir = Path::new("~/lightwhisper/models/parakeet-tdt/");
// Required files:
// - encoder-model.onnx
// - decoder_joint-model.onnx
// - vocab.txt

parakeet.load_model(model_dir)?;

// Transcribe (language parameter ignored - auto-detection)
let samples: &[f32] = /* 16kHz mono audio */;
let text = parakeet.transcribe(samples, "auto")?;
// Supports 25 European languages with automatic detection
```

--------------------------------

### POST save_config

Source: https://context7.com/aluzed/light-whisper/llms.txt

Saves the application configuration to disk and reloads the STT engine if necessary.

```APIDOC
## POST save_config

### Description
Saves the application configuration to disk and reloads the STT engine if the engine type or model size changed.

### Method
POST

### Endpoint
save_config

### Parameters
#### Request Body
- **config** (object) - Required - The configuration object containing audio_device, model_size, language, engine, and shortcut.

### Request Example
{
  "config": {
    "audio_device": "MacBook Pro Microphone",
    "model_size": "small",
    "language": "en",
    "engine": "whisper",
    "shortcut": "Alt+Space"
  }
}
```

--------------------------------

### Listen for Model Download Progress

Source: https://context7.com/aluzed/light-whisper/llms.txt

Listens for 'download-progress' and 'download-complete' IPC events during model downloads. Updates a progress bar and text to show download status.

```javascript
const { event } = window.__TAURI__;

event.listen('download-progress', (e) => {
  const { percent, downloaded_mb, total_mb } = e.payload;
  // percent: 0-100 float
  // downloaded_mb: bytes downloaded in MB
  // total_mb: total file size in MB

  document.getElementById('progress-bar').style.width = `${percent}%`;
  document.getElementById('progress-text').textContent =
    `${downloaded_mb.toFixed(1)} / ${total_mb.toFixed(1)} MB (${percent.toFixed(0)}%)`;
});

event.listen('download-complete', () => {
  console.log('Model download finished');
});
```

--------------------------------

### Application Data Flow

Source: https://github.com/aluzed/light-whisper/blob/main/CLAUDE.md

Visual representation of the recording and transcription process triggered by shortcuts.

```text
Alt+Space → do_toggle_recording() → AudioRecorder.start()
  → cpal captures on dedicated thread → Arc<Mutex<Vec<f32>>>
  → emits "waveform-update" events to frontend

Alt+Space again → AudioRecorder.stop() → thread joins
  → resample to 16kHz → WhisperEngine.transcribe()
  → paste::paste_text() (clipboard + Cmd/Ctrl+V simulation)
```

--------------------------------

### Configure WhisperEngine

Source: https://context7.com/aluzed/light-whisper/llms.txt

Direct interface for Whisper models using whisper-rs. Supports language hints for transcription.

```rust
use crate::stt::WhisperEngine;
use std::path::Path;

let mut whisper = WhisperEngine::new();

// Load GGML model file
whisper.load_model(Path::new("~/lightwhisper/models/ggml-base.bin"))?;

// Transcribe with language hint
let samples: &[f32] = /* 16kHz mono audio */;
let text = whisper.transcribe(samples, "en")?;  // "auto" for auto-detection

// Internally uses:
// - SamplingStrategy::Greedy { best_of: 1 }
// - 4 threads
// - suppress_blank: true
// - single_segment: false
```

--------------------------------

### Record Audio with AudioRecorder

Source: https://context7.com/aluzed/light-whisper/llms.txt

Captures microphone input on a dedicated thread. Requires a tauri::AppHandle for event emission.

```rust
use crate::audio::AudioRecorder;
use tauri::AppHandle;

// Create a new recorder instance
let mut recorder = AudioRecorder::new();

// Start recording on specified device
// Emits "waveform-update" and "recording-started" events
recorder.start("default", app_handle.clone())?;
// Or use specific device name:
// recorder.start("MacBook Pro Microphone", app_handle.clone())?;

// Check if currently recording
if recorder.is_recording() {
    println!("Recording in progress...");
}

// Stop recording and get audio samples
// Returns (samples: Vec<f32>, sample_rate: u32)
let (samples, sample_rate) = recorder.stop()?;
// samples: mono f32 audio at device's native sample rate
// sample_rate: e.g., 44100, 48000

// Resample to 16kHz for STT engines
let samples_16k = crate::audio::resample(&samples, sample_rate, 16000);
```

--------------------------------

### List Available Audio Input Devices

Source: https://context7.com/aluzed/light-whisper/llms.txt

Retrieves a list of all available audio input devices on the system. This is useful for populating a dropdown menu in the settings to allow users to select their preferred microphone.

```javascript
const { invoke } = window.__TAURI__.core;

// Get all available input devices
const devices = await invoke('list_audio_devices');

// Returns: string[]
// ["MacBook Pro Microphone", "External USB Microphone", "AirPods Pro"]

// Populate a dropdown
const audioDeviceSelect = document.getElementById('audio-device');
audioDeviceSelect.innerHTML = '<option value="default">Default</option>';
devices.forEach(device => {
  const opt = document.createElement('option');
  opt.value = device;
  opt.textContent = device;
  audioDeviceSelect.appendChild(opt);
});
```

--------------------------------

### POST /download_model

Source: https://context7.com/aluzed/light-whisper/llms.txt

Downloads the specified STT model from HuggingFace with streaming progress events emitted during download.

```APIDOC
## POST /download_model

### Description
Downloads the specified STT model from HuggingFace with streaming progress events emitted during download.

### Method
POST

### Endpoint
/download_model

### Parameters
#### Request Body
- **engine** (string) - Required - The STT engine to download the model for (e.g., "whisper", "parakeet").
- **modelSize** (string) - Optional - The size of the model to download (e.g., "tiny", "base", "small", "medium"). This parameter is ignored for the "parakeet" engine.

### Request Example
```json
{
  "engine": "whisper",
  "modelSize": "base"
}
```

### Response
#### Success Response (200)
This endpoint does not return a specific success response body, but emits `download-progress` and `download-complete` IPC events.

#### Error Response
- **400 Bad Request**: If required parameters are missing or invalid.
- **500 Internal Server Error**: If the download fails on the server side.
```

--------------------------------

### Save Application Configuration

Source: https://context7.com/aluzed/light-whisper/llms.txt

Saves new application settings to disk. The STT engine will automatically reload if the engine type or model size has changed. Configuration is persisted to ~/lightwhisper/config.json.

```javascript
const { invoke } = window.__TAURI__.core;

// Save new configuration
await invoke('save_config', {
  config: {
    audio_device: "MacBook Pro Microphone",
    model_size: "small",
    language: "en",
    engine: "whisper",
    shortcut: "Alt+Space"
  }
});

// Engine is automatically reloaded if engine type or model changed
// Config is persisted to ~/lightwhisper/config.json
```

--------------------------------

### POST /change_shortcut

Source: https://context7.com/aluzed/light-whisper/llms.txt

Changes the global hotkey used to toggle voice recording. Unregisters all existing shortcuts and registers the new one.

```APIDOC
## POST /change_shortcut

### Description
Changes the global hotkey used to toggle voice recording. Unregisters all existing shortcuts and registers the new one.

### Method
POST

### Endpoint
/change_shortcut

### Parameters
#### Request Body
- **shortcut** (string) - Required - The new hotkey to set. Supported format: `Modifier+Key` (e.g., "Ctrl+Shift+R", "Alt+Space"). Modifiers include Ctrl, Alt, Shift, Super (Meta/Cmd). Keys include A-Z, 0-9, F1-F12, Space, etc.

### Request Example
```json
{
  "shortcut": "Ctrl+Shift+R"
}
```

### Response
#### Success Response (200)
Indicates the shortcut was successfully changed.

#### Error Response
- **400 Bad Request**: If the provided shortcut format is invalid.
```

--------------------------------

### Link and Image Styling

Source: https://github.com/aluzed/light-whisper/blob/main/docs/index.html

Basic styling for anchor tags to inherit color and remove underlines, and for images to ensure they are responsive within their containers.

```css
a {
  text-decoration: none;
  color: inherit;
}
img {
  max-width: 100%;
}
```

--------------------------------

### Download STT Models

Source: https://context7.com/aluzed/light-whisper/llms.txt

Download STT models (Whisper and Parakeet) from HuggingFace with progress events. Check if models already exist before downloading.

```rust
use crate::model_manager::{
    download_whisper_model,
    download_parakeet_model,
    whisper_model_exists,
    parakeet_model_exists,
    model_exists_for_engine,
};
use tauri::AppHandle;

// Check if models exist
let has_whisper_base = whisper_model_exists("base");
let has_parakeet = parakeet_model_exists();
let has_model = model_exists_for_engine("whisper", "small");

// Download Whisper model (async)
let model_path = download_whisper_model("base", app_handle.clone()).await?;
// Emits: download-progress, download-complete events
// Returns: PathBuf to ~/lightwhisper/models/ggml-base.bin

// Download Parakeet model (async, downloads 3 files)
let model_dir = download_parakeet_model(app_handle.clone()).await?;
// Emits: download-file-info (per file), download-progress, download-complete
// Returns: PathBuf to ~/lightwhisper/models/parakeet-tdt/
```

--------------------------------

### AudioRecorder API

Source: https://context7.com/aluzed/light-whisper/llms.txt

The AudioRecorder module provides thread-safe audio capture from the microphone using cpal.

```APIDOC
## AudioRecorder

### Description
Thread-safe audio recorder that captures microphone input on a dedicated thread using cpal.

### Methods
- **new()**: Creates a new recorder instance.
- **start(device_name: &str, app_handle: AppHandle)**: Starts recording on the specified device. Emits "waveform-update" and "recording-started" events.
- **is_recording()**: Returns true if currently recording.
- **stop()**: Stops recording and returns a tuple of (samples: Vec<f32>, sample_rate: u32).
```

--------------------------------

### Download STT Models

Source: https://context7.com/aluzed/light-whisper/llms.txt

Downloads specified STT models from HuggingFace. Listen for 'download-progress' and 'download-complete' events for status updates. Supports Whisper and Parakeet engines.

```javascript
const { invoke } = window.__TAURI__.core;
const { event } = window.__TAURI__;

// Listen for download progress events
event.listen('download-progress', (e) => {
  const { percent, downloaded_mb, total_mb } = e.payload;
  console.log(`${downloaded_mb.toFixed(1)}/${total_mb.toFixed(1)} MB (${percent.toFixed(0)}%)`);
  progressBar.style.width = `${percent}%`;
});

// Listen for multi-file download info (Parakeet only)
event.listen('download-file-info', (e) => {
  const { file_index, file_count, file_name } = e.payload;
  console.log(`Downloading file ${file_index}/${file_count}: ${file_name}`);
});

event.listen('download-complete', () => {
  console.log('Download complete!');
});

// Download Whisper model
await invoke('download_model', {
  engine: "whisper",
  modelSize: "base"  // Options: "tiny", "base", "small", "medium"
});
// Downloads from: https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-base.bin
// Saves to: ~/lightwhisper/models/ggml-base.bin

// Download Parakeet model (downloads 3 files: encoder, decoder, vocab)
await invoke('download_model', {
  engine: "parakeet",
  modelSize: ""  // modelSize is ignored for Parakeet
});
// Downloads from: https://huggingface.co/istupakov/parakeet-tdt-0.6b-v3-onnx/
// Saves to: ~/lightwhisper/models/parakeet-tdt/
```

--------------------------------

### Paste Utility

Source: https://context7.com/aluzed/light-whisper/llms.txt

Utility for pasting text into the active application.

```APIDOC
## paste_text

### Description
Pastes text by setting clipboard content and simulating Cmd+V (macOS) or Ctrl+V (Windows/Linux).

### Parameters
- **text** (string) - Required - The text to be pasted into the active application.
```

--------------------------------

### POST check_model_exists

Source: https://context7.com/aluzed/light-whisper/llms.txt

Checks whether the specified STT model has been downloaded and is available for use.

```APIDOC
## POST check_model_exists

### Description
Checks whether the specified STT model has been downloaded and is available for use.

### Method
POST

### Endpoint
check_model_exists

### Parameters
#### Request Body
- **engine** (string) - Required - The STT engine name ("whisper" or "parakeet").
- **modelSize** (string) - Required - The model size (e.g., "base"). Ignored for Parakeet.

### Response
#### Success Response (200)
- **exists** (boolean) - True if the model files are present.

### Request Example
{
  "engine": "whisper",
  "modelSize": "base"
}
```

--------------------------------

### Check STT Model Existence

Source: https://context7.com/aluzed/light-whisper/llms.txt

Verifies if a specified STT model (for either Whisper or Parakeet) is downloaded and available for use. For Parakeet, the modelSize parameter is ignored.

```javascript
const { invoke } = window.__TAURI__.core;

// Check if Whisper "base" model exists
const whisperExists = await invoke('check_model_exists', {
  engine: "whisper",
  modelSize: "base"
});
// Returns: boolean - true if ~/lightwhisper/models/ggml-base.bin exists

// Check if Parakeet model exists
const parakeetExists = await invoke('check_model_exists', {
  engine: "parakeet",
  modelSize: ""  // modelSize is ignored for Parakeet
});
// Returns: boolean - true if ~/lightwhisper/models/parakeet-tdt/ contains all required files

if (!whisperExists) {
  console.log("Model not downloaded - showing download button");
}
```

--------------------------------

### Automate Text Pasting

Source: https://context7.com/aluzed/light-whisper/llms.txt

Pastes text by updating the system clipboard and simulating keyboard shortcuts.

```rust
use crate::paste::{paste_text, get_frontmost_pid, activate_pid};

// Paste transcribed text into active application
paste_text("Hello, this is transcribed text")?;

// Internally:
// 1. Saves current clipboard content
// 2. Sets new text to clipboard
// 3. Waits 50ms for clipboard to be ready
// 4. Simulates Cmd+V or Ctrl+V
// 5. Waits 100ms then restores previous clipboard
```

--------------------------------

### Base CSS Reset and Body Styles

Source: https://github.com/aluzed/light-whisper/blob/main/docs/index.html

Applies a CSS reset for consistent box-sizing and sets base styles for the HTML and body, including font, background, and text color, with smooth transitions for theme changes.

```css
*, *::before, *::after {
  margin: 0;
  padding: 0;
  box-sizing: border-box;
}
html {
  scroll-behavior: smooth;
}
body {
  font-family: 'Inter', system-ui, sans-serif;
  background: var(--bg-primary);
  color: var(--text-primary);
  line-height: 1.6;
  transition: background 0.3s, color 0.3s;
}
```

--------------------------------

### CSS Variables for Theming

Source: https://github.com/aluzed/light-whisper/blob/main/docs/index.html

Defines CSS custom properties for color schemes and UI elements, supporting light and dark themes. These variables are essential for dynamic theme switching.

```css
:root {
  --coral: #FF6F61;
  --blue: #0F4C81;
  --yellow: #F5DF4D;
  --mint: #88D8B0;
  --lavender: #B8A9C9;
  --peach: #FFBE76;
  --sky: #7EC8E3;
  --rose: #F7CAC9;
  --shadow-offset: 5px;
  --radius: 14px;
  --radius-sm: 10px;
}
```

```css
[data-theme="light"] {
  --bg-primary: #FAFAF8;
  --bg-secondary: #FFFFFF;
  --bg-hero: #FF6F61;
  --bg-features: #0F4C81;
  --bg-demo: #F5DF4D;
  --bg-engines: #88D8B0;
  --bg-cta: #B8A9C9;
  --text-primary: #1A1A2E;
  --text-secondary: #4A4A68;
  --text-on-coral: #FFFFFF;
  --text-on-blue: #FFFFFF;
  --text-on-yellow:#1A1A2E;
  --text-on-mint: #1A1A2E;
  --text-on-lavender:#1A1A2E;
  --shadow-color: rgba(26, 26, 46, 0.18);
  --shadow-hero: rgba(0,0,0,0.15);
  --card-bg: #FFFFFF;
  --card-shadow: #1A1A2E;
  --nav-bg: rgba(250, 250, 248, 0.92);
  --toggle-bg: #E8E8E4;
  --toggle-icon: #1A1A2E;
}
```

```css
[data-theme="dark"] {
  --bg-primary: #111127;
  --bg-secondary: #1A1A3E;
  --bg-hero: #C04D42;
  --bg-features: #0A3560;
  --bg-demo: #B8A030;
  --bg-engines: #2E7D5B;
  --bg-cta: #6B5B8D;
  --text-primary: #F0F0F5;
  --text-secondary:#B0B0CC;
  --text-on-coral: #FFFFFF;
  --text-on-blue: #E0E8F0;
  --text-on-yellow:#111127;
  --text-on-mint: #111127;
  --text-on-lavender:#F0F0F5;
  --shadow-color: rgba(0, 0, 0, 0.35);
  --shadow-hero: rgba(0,0,0,0.3);
  --card-bg: #1E1E42;
  --card-shadow: #000000;
  --nav-bg: rgba(17, 17, 39, 0.92);
  --toggle-bg: #2A2A50;
  --toggle-icon: #F5DF4D;
}
```

--------------------------------

### STT Engine API

Source: https://context7.com/aluzed/light-whisper/llms.txt

Unified interface for speech-to-text transcription using Whisper or Parakeet backends.

```APIDOC
## SttEngine

### Description
Unified speech-to-text engine supporting both Whisper and Parakeet backends.

### Methods
- **from_engine_name(name: &str)**: Creates engine from config name ("whisper" or "parakeet").
- **load_model(path: &Path)**: Loads the model from the specified path.
- **is_loaded()**: Checks if the model is loaded.
- **transcribe(samples: &[f32], language: &str)**: Transcribes 16kHz mono f32 samples. Language parameter is ignored by Parakeet.
```

--------------------------------

### Listen for Waveform Updates

Source: https://context7.com/aluzed/light-whisper/llms.txt

Listens for 'waveform-update' IPC events emitted during recording. These events provide RMS audio level values for real-time waveform visualization. The RMS values are amplified using a sqrt curve for better display.

```javascript
const { event } = window.__TAURI__;

const bars = new Array(30).fill(0);

event.listen('waveform-update', (e) => {
  const rms = e.payload;  // f32 RMS value (typically 0.002-0.05 for speech)

  // Amplify with sqrt curve for better visualization
  const display = Math.min(1.0, Math.sqrt(rms) * 4);

  // Shift bars left and append new value
  bars.shift();
  bars.push(display);

  // Redraw waveform canvas
  drawWaveform(bars);
});
```

--------------------------------

### Handle Application Errors

Source: https://context7.com/aluzed/light-whisper/llms.txt

Listens for 'app-error' IPC events, which are emitted when errors occur during recording, transcription, or pasting. Displays errors using a toast notification. Common errors include audio detection issues, missing models, and processing failures.

```javascript
const { event } = window.__TAURI__;

event.listen('app-error', (e) => {
  const errorMessage = e.payload;
  // Display toast notification with error message
  showToast(errorMessage);

  // Common errors:
  // - "No audio detected..." - microphone permission or device issue
  // - "STT engine not loaded..." - no model downloaded
  // - "Transcription failed..." - model processing error
  // - "Paste failed..." - accessibility permission needed on macOS
});
```

--------------------------------

### Hero Section Styling

Source: https://github.com/aluzed/light-whisper/blob/main/docs/index.html

Styles the main hero section with a full viewport height, centered content, background gradients, and decorative pseudo-elements. It includes responsive font sizes and button styling.

```css
.hero {
  min-height: 100vh;
  display: flex;
  align-items: center;
  justify-content: center;
  background: var(--bg-hero);
  color: var(--text-on-coral);
  padding-top: 140px;
  position: relative;
  overflow: hidden;
}
.hero::before {
  content: '';
  position: absolute;
  top: -60px;
  right: -60px;
  width: 300px;
  height: 300px;
  background: var(--yellow);
  border-radius: 50%;
  opacity: 0.15;
}
.hero::after {
  content: '';
  position: absolute;
  bottom: -40px;
  left: -40px;
  width: 200px;
  height: 200px;
  background: var(--mint);
  border-radius: 50%;
  opacity: 0.12;
}
.hero-content {
  max-width: 720px;
  text-align: center;
  position: relative;
  z-index: 1;
}
.hero-badge {
  display: inline-block;
  background: rgba(255,255,255,0.2);
  padding: 6px 18px;
  border-radius: 50px;
  font-size: 0.85rem;
  font-weight: 600;
  margin-bottom: 28px;
  backdrop-filter: blur(4px);
}
.hero h1 {
  font-size: clamp(2.8rem, 6vw, 4.5rem);
  font-weight: 800;
  line-height: 1.1;
  margin-bottom: 20px;
}
.hero p {
  font-size: clamp(1.05rem, 2vw, 1.25rem);
  opacity: 0.92;
  max-width: 520px;
  margin: 0 auto 40px;
  line-height: 1.6;
}
.hero-buttons {
  display: flex;
  gap: 16px;
  justify-content: center;
  flex-wrap: wrap;
}
.btn {
  display: inline-flex;
  align-items: center;
  gap: 10px;
  paddin
```

--------------------------------

### Section Base Padding

Source: https://github.com/aluzed/light-whisper/blob/main/docs/index.html

Sets consistent top and bottom padding for all section elements to ensure vertical spacing across the page.

```css
section {
  padding: 100px 40px;
}
```

--------------------------------

### Resample Audio Data

Source: https://context7.com/aluzed/light-whisper/llms.txt

Performs linear interpolation to convert audio to 16kHz, which is required for STT engines.

```rust
use crate::audio::resample;

// Resample from 48kHz to 16kHz (required for Whisper/Parakeet)
let samples_48k: Vec<f32> = /* audio at 48kHz */;
let samples_16k = resample(&samples_48k, 48000, 16000);

// No-op if rates match
let same = resample(&samples_48k, 48000, 48000);  // Returns clone
```

--------------------------------

### IPC Events

Source: https://context7.com/aluzed/light-whisper/llms.txt

Real-time events emitted from the backend to the frontend for various application states and updates.

```APIDOC
## IPC Events (Backend to Frontend)

### `waveform-update`

#### Description
Emitted during recording with RMS audio level values for real-time waveform visualization.

#### Payload
- **rms** (f32) - The Root Mean Square audio level value.

### `recording-started` / `recording-stopped`

#### Description
Emitted when recording begins or ends to update UI state (timers, waveform reset).

#### Payload
None.

### `download-progress` / `download-complete`

#### Description
Emitted during model downloads with progress information.

#### Payload (`download-progress`)
- **percent** (f32) - Download progress percentage (0-100).
- **downloaded_mb** (f32) - Amount of data downloaded in megabytes.
- **total_mb** (f32) - Total file size in megabytes.

#### Payload (`download-complete`)
None.

### `app-error`

#### Description
Emitted when errors occur (recording failed, transcription failed, paste failed).

#### Payload
- **errorMessage** (string) - A message describing the error that occurred.
```

--------------------------------

### Navigation Bar Styling

Source: https://github.com/aluzed/light-whisper/blob/main/docs/index.html

Styles the fixed navigation bar with background blur effects, logo display, navigation links, and a theme toggle button. Includes hover and active states for interactive elements.

```css
nav {
  position: fixed;
  top: 0;
  left: 0;
  right: 0;
  z-index: 100;
  display: flex;
  align-items: center;
  justify-content: space-between;
  padding: 16px 40px;
  background: var(--nav-bg);
  backdrop-filter: blur(12px);
  -webkit-backdrop-filter: blur(12px);
  transition: background 0.3s;
}
.nav-logo {
  font-size: 1.2rem;
  font-weight: 800;
  display: flex;
  align-items: center;
  gap: 8px;
}
.nav-logo svg {
  width: 28px;
  height: 28px;
}
.nav-links {
  display: flex;
  align-items: center;
  gap: 24px;
}
.nav-links a {
  font-size: 0.9rem;
  font-weight: 600;
  opacity: 0.8;
  transition: opacity 0.2s;
}
.nav-links a:hover {
  opacity: 1;
}
.theme-toggle {
  width: 42px;
  height: 42px;
  border-radius: 50%;
  border: none;
  cursor: pointer;
  background: var(--toggle-bg);
  color: var(--toggle-icon);
  display: flex;
  align-items: center;
  justify-content: center;
  font-size: 1.2rem;
  transition: background 0.3s, color 0.3s, transform 0.2s;
  box-shadow: 3px 3px 0 var(--shadow-color);
}
.theme-toggle:hover {
  transform: translateY(-2px);
}
.theme-toggle:active {
  transform: translate(2px, 2px);
  box-shadow: none;
}
```

--------------------------------

### Language Toggle Button Styling

Source: https://github.com/aluzed/light-whisper/blob/main/docs/index.html

Styles a language toggle component with pill-shaped buttons and active state highlighting. It uses background and text color transitions for a smooth user experience.

```css
.lang-toggle {
  display: flex;
  align-items: center;
  gap: 6px;
  background: var(--toggle-bg);
  border-radius: 50px;
  padding: 4px;
  box-shadow: 3px 3px 0 var(--shadow-color);
  transition: background 0.3s;
}
.lang-toggle button {
  display: flex;
  align-items: center;
  gap: 4px;
  padding: 5px 12px;
  border: none;
  border-radius: 50px;
  font-size: 0.82rem;
  font-weight: 700;
  cursor: pointer;
  background: transparent;
  color: var(--text-secondary);
  transition: background 0.2s, color 0.2s, box-shadow 0.2s;
}
.lang-toggle button.active {
  background: var(--coral);
  color: #fff;
  box-shadow: 2px 2px 0 rgba(0,0,0,0.12);
}
.lang-toggle button:not(.active):hover {
  color: var(--text-primary);
}
```

--------------------------------

### Change Global Hotkey

Source: https://context7.com/aluzed/light-whisper/llms.txt

Changes the global hotkey for toggling voice recording. This function unregisters all existing shortcuts and registers the new one. Supported format: Modifier+Key.

```javascript
const { invoke } = window.__TAURI__.core;

// Change hotkey to Ctrl+Shift+R
try {
  await invoke('change_shortcut', { shortcut: "Ctrl+Shift+R" });
  console.log("Shortcut updated successfully");
} catch (e) {
  console.error(`Invalid shortcut: ${e}`);
}

// Reset to default
await invoke('change_shortcut', { shortcut: "Alt+Space" });

// Supported format: Modifier+Key
// Modifiers: Ctrl, Alt, Shift, Super (Meta/Cmd)
// Keys: A-Z, 0-9, F1-F12, Space, etc.
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.