Wujian Speech AI Platform

T-head's Wujian Speech AI Platform utilizes a heterogeneous multi-core architecture, supplemented by a speech algorithm hardware accelerator and multilevel low-power consumption management scheme, to help customers quickly build a lightweight, terminal-side artificial intelligence speech product with ultra-low power consumption and ultra-low cost.

Architecture Features

  • AI enhanced CPU: 128-bit SIMD architecture, single-cycle parallel processing of 16 8-bit operations

  • Heterogeneous multi-core architecture: Dual lightweight processor, with MCU control, speech codec and AI computing

  • Speech algorithm accelerator: Contains basic units such as MAC, EXP, VEC, FIR and FFT/IFFT

  • Multilevel low-power management: Includes hardware VAD, low-voltage retention, and multiple-power domain management

  • Complete audio interface: Applicable to linear array, ring array and built-in AEC loop from one to eight microphones

  • Speech algorithm support: Far field pickup enhancement algorithm, and local keyword recognition algorithm

Featured Technology

  • Characteristics of technology

Industrial Applications