Metadata-Version: 2.4
Name: voxing
Version: 0.1.1
Summary: Voice assistant TUI powered by local MLX models
Author: yeungadrian
License-Expression: MIT
Classifier: Development Status :: 3 - Alpha
Classifier: Environment :: Console
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3.13
Classifier: Operating System :: MacOS
Requires-Dist: langdetect>=1.0.9
Requires-Dist: mlx-audio[tts]==0.3.1
Requires-Dist: mlx-lm>=0.30.5
Requires-Dist: pip>=26.0.1
Requires-Dist: pydantic-settings>=2.12.0
Requires-Dist: rich>=14.3.1
Requires-Dist: sounddevice>=0.5.3
Requires-Dist: textual>=7.5.0
Requires-Dist: typer>=0.23.1
Requires-Python: ==3.13.*
Project-URL: Homepage, https://github.com/yeungadrian/Voxing
Project-URL: Repository, https://github.com/yeungadrian/Voxing
Description-Content-Type: text/markdown

# Voxing

A voice and text assistant that runs entirely on-device, built as a terminal UI for macOS with Apple Silicon.

Uses [MLX](https://github.com/ml-explore/mlx) for inference and [Textual](https://github.com/Textualize/textual) for the interface.

## Requirements

- macOS with Apple Silicon (M1+)
- 16 GB RAM (app uses ~4-5 GB)

## Install

```bash
uvx voxing
```

## Models

| Task | Default Model |
|------|---------------|
| Speech-to-Text | [parakeet-tdt-0.6b-v3](https://huggingface.co/mlx-community/parakeet-tdt-0.6b-v3) |
| LLM | [LFM2.5-1.2B-Instruct-MLX-8bit](https://huggingface.co/LiquidAI/LFM2.5-1.2B-Instruct-MLX-8bit) |
| Text-to-Speech | [Kokoro-82M-bf16](https://huggingface.co/mlx-community/Kokoro-82M-bf16) |

Models can be swapped at runtime with the `/model` command.

## Usage

| Command | Description |
|---------|-------------|
| `/record` | Record voice input (stops on silence) |
| `/transcribe` | Extended recording (up to 3 min), copies to clipboard |
| `/model` | Switch STT, LLM, or TTS model |
| `/tts` | Toggle text-to-speech |
| `Esc` (x2) | Cancel active operation |
