Metadata-Version: 2.4
Name: suzune
Version: 0.1.0
Summary: Universal On-Device Speech AI Framework by Sorika Labs
Author-email: Sorika Labs <contact@sorikalabs.com>
Project-URL: Homepage, https://huggingface.co/sorika-labs
Project-URL: Repository, https://huggingface.co/sorika-labs
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: Apache Software License
Classifier: Operating System :: OS Independent
Classifier: Topic :: Multimedia :: Sound/Audio :: Speech
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Requires-Python: >=3.9
Description-Content-Type: text/markdown
Requires-Dist: suzune-flash>=0.1.0
Requires-Dist: suzune-mini>=0.1.0

# 🌸 Suzune: Universal On-Device Speech AI Framework

[![PyPI version](https://img.shields.io/badge/pypi-v0.1.0-blue.svg)](https://pypi.org/project/suzune/)
[![Hugging Face](https://img.shields.io/badge/HF%20Org-sorika--labs-yellow)](https://huggingface.co/sorika-labs)
[![License](https://img.shields.io/badge/License-Apache%202.0-green.svg)](LICENSE)

**Suzune** is the unified Python SDK for **Sorika Labs'** foundation text-to-speech models.

It provides a single, intuitive interface to run both **Suzune Flash** (82M flagship edge) and **Suzune Mini** (40M causal flow) on-device with zero cloud latency.

---

## 🚀 Quick Install

```bash
# Install the unified SDK
pip install suzune

# Or install individual models
pip install suzune-flash   # 82M Multilingual Flagship (Ira, Kian, Kagami, Phi)
pip install suzune-mini    # 40M Ultra-Fast Causal Flow (<10ms latency)
```

---

## 🎙️ Python Quickstart

```python
import suzune

# ⚡ Option A: Load Suzune Flash (Flagship Edge)
flash = suzune.load("flash", voice="ira")
flash.speak("Hello! Suzune Flash delivers studio-quality speech on device.", output="flash.wav")

# 🌸 Option B: Load Suzune Mini (Ultra-Fast 7ms Causal Flow)
mini = suzune.load("mini")
mini.speak("Hello! Suzune Mini is optimized for real-time edge devices.", output="mini.wav")
```

---

## 🎭 Models at a Glance

| Model | Parameters | Latency | Key Strengths |
| :--- | :--- | :--- | :--- |
| **Suzune Flash** | 82M | ~20ms | Multilingual, studio fidelity, signature voices (`ira`, `kian`, `kagami`, `phi`), GGUF support |
| **Suzune Mini** | 40M | <10ms | Ultra-lightweight causal flow, sub-10ms time-to-first-audio, ideal for embedded IoT & robotics |

---

## 🏛️ About Sorika Labs
Sorika Labs is an independent AI research organization dedicated to developing ultra-compact, privacy-first foundation models for edge and on-device intelligence.
