ChivoxAI

/docs

VAD Document

1. Description

Vad is used to detect the start and stop states of user's speech in real time, and the application layer makes corresponding prompts and processing according to different states returned to improve user experience.

1.1 Audio data source

Microphone live audio streaming

1.2 Return data description

1.2.1 Data type

json format

1.2.2 Field Description
Name Type Description
vad_status int Detection state, the values are 0,1,2
0: Start detection;
1: Detected that the user started to speak;
2: Detected that the user has stopped speaking;
sound_intensity int Sound intensity, value range:0~100
1.2.3 Data sample
    {
	   "vad_status": 0,
	   "sound_intensity": 12.0
    }

2. Integrated

VAD function can be integrated in the following two ways according to different solutions

Type Applicable Scenarios Supported Platforms/Development Languages Doc reference
Integrated in the evaluation sdk Products using the evaluation sdk solution Android、iOS、Unity、C/C++、Java、C# Click to view
Separate vad sdk Products using API solutions Android、iOS Click to view

SDK, API, MCP and Function Calling documentation on this site.