Skip to content

CI13XX Series Chip SDK

The CI13XX series chips currently have the following SDK versions categorized by different application scenarios:

1. CI-SDK-ASR-Offline

Offline Speech Recognition SDK

Note

SDK Release Version Format: CI13XX_SDK_ASR_Offline_Vx.x.x

Primarily designed for pure offline simple application scenarios, such as smart home applications. Supported audio front-end algorithms include:

  • ASR (Automatic Speech Recognition - Single Microphone)
  • CWSL (Offline Commands Self-Learning )
  • AEC (Acoustic Echo Cancellation)

2. CI-SDK-ASR-ALG

Offline Voice Algorithm SDK

Note

SDK Release Version Format: CI13XX_SDK_ASR_ALG_Vx.x.x

The CI13XX series chips integrate the BNPU V3 with strong neural network and vector parallel computing capabilities, along with sufficient RAM, enabling the implementation of various common speech processing algorithms. The main algorithm features include the following. Users can make appropriate selections based on different product requirements.

  • ASR (Automatic Speech Recognition - Single Microphone, fixed entries or natural speech)
  • CWSL (Offline Commands Self-Learning )
  • AEC (Acoustic Echo Cancellation)
  • Voiceprint Recognition: Includes VPR (Voiceprint Recognition based on specific individuals), WMAN_VPR (Gender Detection), etc.
  • Sound Event Detection: SED_CRY (Crying Detection), SED_SNORE (Snoring Detection), Cough Detection, etc.
  • DENOISE_NN (Neural Network-based Deep Noise Reduction)
  • DOA (Dual-Microphone Direction of Arrival)
  • DERVERB (Dual-Microphone Dereverberation)
  • BF (Beamforming - Dual-Microphone Deep Separation)
  • PWK (Power Value Calculation, distinguishing target sound distance)
  • ALC (Automatic Level Control)
  • TTS (Text-to-Speech - Supports Chinese, numbers, and letters only, does not support English)

The ASR feature can be combined in a 1+1 configuration with other features except TTS. For multiple algorithm combinations, please refer to ☞Noise Reduction Development Guide.

3. CI-SDK-NN-ENC

Voice Call Noise Reduction SDK

Note

SDK Release Version Format: CI13XX_SDK_NN_ENC_Vx.x.x

Under Development…

4. CI-SDK-LLM-AIOT

Offline/Online Large Model Dialogue SDK

Note

SDK Release Version Format: CI13XX_SDK_LLM_AIOT_Vx.x.x

Under Development…

Note

All SDK versions can be downloaded from the development resources section at ☞ChipIntelli AI Speech Development Platform.