Shadowgraph
About
Sign in

superxiaobai-1/LLM_Voice_Flow

100

stars

1

commits

C++

primary language

Aug 17, 2025

updated

README

端侧部署C++项目-基于DeepSeek与RK3576的模块化离线智能语音交互系统

项目视频文档解析

添加微信：auto_drive_yue

项目概述

本项目开发了一套全离线、模块化的智能语音交互系统，基于RK3576 NPU实现端到端智能语音交互流水线，集成流式ASR、DeepSeek大模型推理、TTS语音合成-双缓冲队列三大核心模块。系统采用松耦合架构，各模块通过标准化接口(封装ZeroMQ通信协议)交互，在嵌入式环境下实现4秒内的语音输入→LLM思考→语音输出闭环。

项目技术栈

技术栈：Linux、C++、ASR、RK芯片云端量化/端侧部署、TTS、ZeroMQ、CMake、多线程

项目架构

System Architecture

核心特性

  • 🚀 全离线部署：不依赖云端服务，基于RK3576 NPU实现本地化推理
  • 🔧 模块化架构：ASR/TTS/LLM模块通过ZeroMQ松耦合通信
  • ⚡ 低延迟优化：流式ASR + 双缓冲TTS队列实现快速响应

Contributors

18362576243

1 commits

Similar repos

100askTeam/xiaozhi-linux

Embedded Linux AI Xiaozhi Intelligent voice dialogue.

282

GPAUP/nlp_team

0

soniqo/speech-core

On-device VAD / streaming STT / TTS / diarization in C++17 (ONNX + LiteRT) with a voice-agent…

78

hackers365/xiaozhi-esp32-server-golang

golang版本的小智后端服务，支持websocket和mqtt+udp协议，支持声纹识别/声音克隆/知识库/mcp远程调用/主动音频下发/openclaw等功能

411

guokezhen999/VoiceHub

Cross-platform, on-device voice AI app for speech recognition, translation, simultaneous…

0

tc-mb/llama.cpp-omni

Omni inference in C/C++

258

asiff00/On-Device-Speech-to-Speech-Conversational-AI

This is an on-CPU real-time conversational system for two-way speech communication with AI models,…

257

gkintu/hf-speech-to-speech

A fully local realtime voice agent using Qwen3.5, Parakeet TDT, and Qwen3-TTS

5

Languages

C++

79.0%

Fortran

3.9%

Python

2.9%

C

2.5%

CMake

2.0%

Makefile

1.7%

Shell

1.1%

JavaScript

1.0%

superxiaobai-1/LLM_Voice_Flow

100

stars

1

commits

C++

primary language

Aug 17, 2025

updated

README

端侧部署C++项目-基于DeepSeek与RK3576的模块化离线智能语音交互系统

项目视频文档解析

添加微信：auto_drive_yue

项目概述

本项目开发了一套全离线、模块化的智能语音交互系统，基于RK3576 NPU实现端到端智能语音交互流水线，集成流式ASR、DeepSeek大模型推理、TTS语音合成-双缓冲队列三大核心模块。系统采用松耦合架构，各模块通过标准化接口(封装ZeroMQ通信协议)交互，在嵌入式环境下实现4秒内的语音输入→LLM思考→语音输出闭环。

项目技术栈

技术栈：Linux、C++、ASR、RK芯片云端量化/端侧部署、TTS、ZeroMQ、CMake、多线程

项目架构

System Architecture

核心特性

  • 🚀 全离线部署：不依赖云端服务，基于RK3576 NPU实现本地化推理
  • 🔧 模块化架构：ASR/TTS/LLM模块通过ZeroMQ松耦合通信
  • ⚡ 低延迟优化：流式ASR + 双缓冲TTS队列实现快速响应

Similar repos

100askTeam/xiaozhi-linux

Embedded Linux AI Xiaozhi Intelligent voice dialogue.

282

GPAUP/nlp_team

0

soniqo/speech-core

On-device VAD / streaming STT / TTS / diarization in C++17 (ONNX + LiteRT) with a voice-agent…

78

hackers365/xiaozhi-esp32-server-golang

golang版本的小智后端服务，支持websocket和mqtt+udp协议，支持声纹识别/声音克隆/知识库/mcp远程调用/主动音频下发/openclaw等功能

411

guokezhen999/VoiceHub

Cross-platform, on-device voice AI app for speech recognition, translation, simultaneous…

0

tc-mb/llama.cpp-omni

Omni inference in C/C++

258

asiff00/On-Device-Speech-to-Speech-Conversational-AI

This is an on-CPU real-time conversational system for two-way speech communication with AI models,…

257

gkintu/hf-speech-to-speech

A fully local realtime voice agent using Qwen3.5, Parakeet TDT, and Qwen3-TTS

5

Contributors

18362576243

1 commits

Languages

C++

79.0%

Fortran

3.9%

Python

2.9%

C

2.5%

CMake

2.0%

Makefile

1.7%

Shell

1.1%

JavaScript

1.0%