93 lines
2.1 KiB
Markdown
93 lines
2.1 KiB
Markdown
# Installation Guide
|
|
|
|
This guide provides instructions for installing and setting up Ubuntu Voice Typing.
|
|
|
|
## Requirements
|
|
|
|
- Ubuntu 22.04 or newer
|
|
- Python 3.6 or newer
|
|
- Dependencies for speech recognition and text injection
|
|
|
|
## Basic Installation
|
|
|
|
### 1. Install System Dependencies
|
|
|
|
```bash
|
|
# For X11 environments
|
|
sudo apt install xdotool python3-pip python3-gi gir1.2-appindicator3-0.1
|
|
|
|
# For Wayland environments
|
|
sudo apt install wtype python3-pip python3-gi gir1.2-appindicator3-0.1
|
|
# OR alternatively for Wayland
|
|
sudo apt install ydotool python3-pip python3-gi gir1.2-appindicator3-0.1
|
|
```
|
|
|
|
### 2. Clone the Repository
|
|
|
|
```bash
|
|
git clone https://github.com/ubuntu-voice-typing/ubuntu-voice-typing.git
|
|
cd ubuntu-voice-typing
|
|
```
|
|
|
|
### 3. Install Python Dependencies
|
|
|
|
```bash
|
|
# Basic installation
|
|
pip3 install --user .
|
|
|
|
# With Whisper support (requires more resources)
|
|
pip3 install --user ".[whisper]"
|
|
|
|
# For development
|
|
pip3 install --user ".[dev]"
|
|
```
|
|
|
|
### 4. Run the Application
|
|
|
|
```bash
|
|
# Basic usage
|
|
ubuntu-voice-typing
|
|
|
|
# With debug logging
|
|
ubuntu-voice-typing --debug
|
|
|
|
# With Whisper instead of VOSK
|
|
ubuntu-voice-typing --engine whisper
|
|
|
|
# Force Wayland mode
|
|
ubuntu-voice-typing --wayland
|
|
```
|
|
|
|
## Advanced Installation
|
|
|
|
### Custom Model Sizes
|
|
|
|
By default, a small model will be downloaded and used. You can use larger models for improved accuracy:
|
|
|
|
```bash
|
|
ubuntu-voice-typing --model medium
|
|
# or
|
|
ubuntu-voice-typing --model large
|
|
```
|
|
|
|
Note that larger models require more RAM and may have longer initialization times, but provide better recognition accuracy.
|
|
|
|
## Troubleshooting
|
|
|
|
### No audio input detected
|
|
|
|
- Check that your microphone is connected and working
|
|
- Verify your system audio settings
|
|
- Run with `--debug` flag to see detailed logs
|
|
|
|
### Text injection not working
|
|
|
|
- For X11: Ensure xdotool is installed
|
|
- For Wayland: Ensure wtype or ydotool is installed
|
|
- Some applications may have security measures that prevent text injection
|
|
|
|
### Recognition is inaccurate
|
|
|
|
- Try a larger model size (`--model medium` or `--model large`)
|
|
- Ensure your microphone is positioned correctly
|
|
- Speak clearly and at a moderate pace |