Products

CNSV

CNSV, a speaker verification/identification SDK, utilizes neural network models to deliver accurate performance while being optimized for edge computing on MCU platforms, making it ideal for IoT and low-power devices.

Features

  • Text-Independent

    • No constraint on speech content.

  • High performance

    • Low recognition error rates & high robustness by NN model.
    • Utilizing Arm CMSIS-NN to enhance computation efficiency on Cortex-M MCU.

  • Supporting Speaker Identification

    • Enables identification among multiple registered speakers, allowing systems to recognize and distinguish different users' voices.

Solutions

Local speaker verification in smart speaker, TWS, car system, IoT and smart appliances.