百聆

百聆voice conversation assistant

百聆is an open source voice conversation assistant that provides a natural and smooth voice interaction experience. It combines speech recognition, voice activity detection, large language model and speech synthesis technology.

Main advantages

Achieving dialogue effects close to GPT-4 without the need for a GPU

Suitable for a variety of edge devices and resource-constrained environments

Completely open source, supporting community contributions and secondary development

Efficient and reliable, providing high-quality voice dialogue

target users

Users who need an efficient voice interaction experience, such as daily conversations, information query or task management. It is especially suitable for users who do not have high requirements for hardware configuration but have high requirements for voice interaction quality.

Usage scenario examples

Voice query weather

IELTS Speaking Practice

Set up scheduled tasks

Product features

Efficient open source model

Deploy locally, no GPU required

Modular design for easy customization and upgrade

Support memory function to provide personalized experience

Support tool calling to improve practicality

Support task management and efficiently manage user tasks

Tutorial

1 Clone the project warehouse git clone https://github.com/wwbin2017/bailing.git and enter the project directory

2 Install dependencies pip install -r requirements.txt

3 Configure the environment variable configuration config/config.yaml, download SenseVoiceSmall to the models/SenseVoiceSmall directory, and obtain the api_key of deepseek

4 Start the backend service (optional) cd server and run python server.py

5 Start the main program and run python main.py to start the voice conversation