3.1.Project Overview
What is tkvoice?
tkvoice is a complete offline speech interaction system based on TienKung platform, integrating three core functionalities:
| Function | Abbreviation | Purpose | Analogy |
|---|---|---|---|
| Automatic Speech Recognition | ASR | Convert user's speech to text | Converting speech to typing |
| Large Language Model | LLM | Understand text, generate answer text | Understanding questions and thinking |
| Text-to-Speech | TTS | Convert text to machine speech | Converting text to speech |
Complete Speech Interaction Flow
User speaks
↓
[ASR] Speech Recognition → Get text
↓
[LLM] Large Language Model → Generate answer
↓
[TTS] Text-to-Speech → Generate speech
↓
Machine speaks
Why Use Offline Solution?
- ✅ Privacy Protection - Data stays local, no cloud transmission
- ✅ Low Latency - Unaffected by network, real-time interaction
- ✅ High Reliability - Network-independent, no cloud service accounts required
- ✅ Cost-Effective - One-time deployment, no service fees
- ⚠️ High Resource Requirements - Requires strong computing power