This project provides an STM32 Microcontroler embedded real time environement to execute X-CUBE-AI generated model targetting audio applications. The purpose of this package is to stream physical data acquired by sensors into a processing chain including a preprocessing step that typically would perform a first level of feature extraction, the machine learning inference itself, and a post processing step before exposing the results to the user in real time.
- Deployement through ST Model Zoo
- Support of Audio Event Detection (AED) using Yamnet 1024
- Low power support:
- Dynamic Voltage & Frequency Scaling (DFVS)
- Dynamic Power Scaling (DPS)
- Bare Metal (BM) Implementation
- Real Time Operating System (RTOS) Implementation
- Random load generator
- Audio bypass
- Audio loop back
| Name | Version | Release notes |
|---|---|---|
| STEdge AI runtime | v4.0.0 | |
| STM32 AI AudioPreprocessing Library | v1.2.1 | release notes |
| FreeRTOS kernel | v10.6.2 | release notes |
| CMSIS | v5.9.0_fix_620_dsp_v1.10.0_nn_v3.1.0 | release notes |
| STM32N6xx CMSIS Device | v1.1.0 | release notes |
| STM32N6xx HAL/LL Drivers | v1.1.0 | release notes |
| STM32N6570-DK BSP Drivers | v1.1.0 | release notes |
| BSP Component aps256xx | v1.0.6 | release notes |
| BSP Component cs42l51 | v2.0.6 | release notes |
| BSP Component mx66uw1g45g | v1.1.0 | release notes |
| BSP Component wm8904 | v1.1.0 | release notes |
| BSP Component Common | v7.3.0 | release notes |
- Upgrade to STEdgeAI 4.0.0
- Upgrade to STM32CUBEIDE V 2.1.0
- Upgrade to STEdgeAI 3.0.0:
- Migrate to new APIs
- Migrate ThreadX project to freertos project
- Upgrade to STEdgeAI 2.2.0
- upgraded to
- Cube Firmware N6 V 1.1
- STEdge AI V 2.1
- STM32CUBEIDE V 1.18.1
- Support of Audio Event Detection (AED) using Yamnet 1024
- Low power support:
- Dynamic Voltage & Frequency Scaling (DFVS)
- Dynamic Power Scaling (DPS)
- Bare Metal (BM) Implementation
- Real Time Operating System (RTOS) Implementation
- Audio loop Back
- OTP management