dictation-input

KO | EN

gitlite — search

#python #hacktoberfest #llm #golang #nodejs #macos #awesome #vue #awesome-list #pytorch #security #mcp #openai #open-source

dictation-input

(★ 11)

Press to speak, speak to input

dictation-input Latest Version Download

Download Latest Version (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Linux Dictation Overlay

A beautiful, high-performance, Siri-like voice dictation overlay for Linux (Wayland/Hyprland).

This project provides a visual feedback layer for your voice dictation workflow. It features a fluid, audio-reactive glowing orb animation while recording, and a minimal pulsing indicator while transcribing.

https://github.com/user-attachments/assets/209673d4-4012-430f-9c8c-937c5c3d8f9f

## Features

- **Audio Reactive:** The recording overlay pulses and expands in real-time based on your voice volume (zero latency).
- **Beautiful Animations:**
  - **Recording:** "Siri-like" multi-colored orbiting glowing orbs.
  - **Transcribing:** Minimalist white pulsing dots.
- **Wayland Native:** Built with `gtk-layer-shell` for seamless integration with compositors like Hyprland and Sway.
- **Fast & Lightweight:** Written in C with Cairo graphics for 60FPS smooth animations on minimal resources.
- **Auto-Stop:** Automatically stops and transcribes if you forget to stop recording (1-minute timeout).

## Requirements

You need a Linux system (preferably Wayland) with the following dependencies:

- **GTK3**
- **gtk-layer-shell**
- **PulseAudio (libpulse-simple)**
- **Whisper.cpp** (for the actual speech-to-text)
- **pw-record** (PipeWire tools) or similar recorder
- **wtype** (for simulating keypresses to insert text)

### Install Dependencies (Arch Linux)

```bash
sudo pacman -S gtk3 gtk-layer-shell libpulse pipewire whisper.cpp wtype
```

*(Note: If using a different distro, adjust package names accordingly).*

## Installation

1.  **Clone the repository:**

```bash
    git clone https://github.com/yourusername/dictation-overlay.git
    cd dictation-overlay
    ```

2.  **Compile the overlays:**

```bash
    make
    ```

This will create two binaries: `dictation-overlay` (recording) and `transcribe-overlay` (loading).

3.  **Set up the script:**

Ensure `dictation.sh` is executable and points to your paths.

```bash
    chmod +x dictation.sh
    ```

**Important:** Open `dictation.sh` and update the following variables to match your system:

```bash
    # Path to your compiled binaries
    RECORD_APP="$HOME/code/personal/dictation-overlay/dictation-overlay"
    TRANSCRIBE_APP="$HOME/code/personal/dictation-overlay/transcribe-overlay"
    
    # Path to your Whisper model
    MODEL="$HOME/.whisper/models/ggml-base-q5_1.bin" 
    ```

## Usage

The intended workflow is "Push-to-Talk" or "Toggle-to-Talk".

### Manual Test

```bash
./dictation.sh start
# Speak now...
./dictation.sh stop
# Watch it transcribe and type!
```

### Hyprland Configuration

Bind a key (e.g., `Super + D`) to start recording when held, and stop when released.

**Add this to `~/.config/hypr/hyprland.conf`:**

```conf
# Hold Super+D to dictate
bind = SUPER, D, exec, /path/to/dictation-overlay/dictation.sh start
bindr = SUPER, D, exec, /path/to/dictation-overlay/dictation.sh stop
```

## Customization

### Adjusting Sensitivity
The audio reactivity is tuned for a standard microphone input. If the bubbles are too small or too large, edit `overlay.c` around line 105:

```c
// Increase/decrease this multiplier
double final_level = rms * 20.0; 
```

Recompile with `make` after changes.

### Window Position
By default, the overlay appears at the **bottom-center** with a 40px margin. To change this, edit the `gtk_layer_set_anchor` lines in `overlay.c`.

## Troubleshooting

- **"Error: Audio file missing":** Check if `pw-record` is installed and working.
- **Overlay doesn't animate:** Ensure your microphone is the default input device in PulseAudio/PipeWire.
- **Overlay is invisible:** Ensure your compositor supports `gtk-layer-shell` and transparency (Hyprland does by default).

## License

MIT License

dictation-input

dictation-input Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

dictation-input

dictation-input Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

Download README