Tai 1.0.0

Tai 1.0.0 — Voice Assistant Core

Tai 1.0.0 establishes the first complete local voice assistant pipeline.

The system can listen continuously, detect user speech, transcribe audio locally, generate responses through a local LLM, synthesize speech locally, and handle user barge-in through an event-driven orchestrator.

Main features

• Continuous voice input
• Local Whisper-based STT
• GPU-accelerated transcription
• Event-driven orchestration
• Local Ollama-backed LLM generation
• Local Piper-based TTS playback
• Barge-in support
• Conversation turn management
• Performance metrics aggregation
• System health debug aggregation
• Modular Java/Python service architecture

Release scope

This release is the foundation of Tai's core.

It does not represent the final multimodal vision yet. UI, avatar, screen vision and long-term memory will be developed in later versions.