320
D. Zhu et al.
2 Project Description
Different from traditional manual input commands, the intelligent voice control
system uses voice information to execute commands. Our system has following
functions:
a. Acquire voice message.
b. Recognizing the command form speech data.
c. Speaker recognition and authentication.
d. Broadcast current status and parameters with synthesized speech when asked.
The platform is divided into three layers: physical layer, data and information
layer, and business logic layer (see Fig. 1).
Fig. 1. System structure
Physical layer consists of sound card adapter, task computer, and interface
module. Sound card adapter includes a chip microprocessor and a sound card,
and it can collect speech signal and play synthesis speech. As for task computer,
this is the main part of speech recognition, speaker authentication, instruction
analysis, and speech synthesis; interface module realize interaction function to
sensors.
Data and information layer includes deep learning framework, configuration
management, health management, and data communication middleware.
Business logic layer consists of voice activity detection, auto speech recognition, speaker authentication, instruction analysis, and automatic answer.
Figure 2 is flow chart of system.
D. Zhu et al.
2 Project Description
Different from traditional manual input commands, the intelligent voice control
system uses voice information to execute commands. Our system has following
functions:
a. Acquire voice message.
b. Recognizing the command form speech data.
c. Speaker recognition and authentication.
d. Broadcast current status and parameters with synthesized speech when asked.
The platform is divided into three layers: physical layer, data and information
layer, and business logic layer (see Fig. 1).
Fig. 1. System structure
Physical layer consists of sound card adapter, task computer, and interface
module. Sound card adapter includes a chip microprocessor and a sound card,
and it can collect speech signal and play synthesis speech. As for task computer,
this is the main part of speech recognition, speaker authentication, instruction
analysis, and speech synthesis; interface module realize interaction function to
sensors.
Data and information layer includes deep learning framework, configuration
management, health management, and data communication middleware.
Business logic layer consists of voice activity detection, auto speech recognition, speaker authentication, instruction analysis, and automatic answer.
Figure 2 is flow chart of system.