The osxAcousticSL software lets you implement a real-time sound source localization algorithm using two or four signals acquired from digital MEMS microphones to estimate the arrival direction of an audio source.
It is based on three different DOA algorithms exploiting cross correlation in the time domain, generalized cross correlation with phase transform and a matching pursuit routine using sparse representation framework.
The angle can be estimated over a 180 or 360 degree range, depending on the number of channels adopted and microphone placement.
The resolution of the computed value can be chosen at runtime, allowing you to determine the best tradeoff between localization precision and resource consumption.
The osxAcousticSL library is provided in binary format inside a software package with sample applications running on the X-NUCLEO-CCA02M1 expansion board connected to a NUCLEO-F401RE board.
The example package is an add-on for the X-CUBE-MEMSMIC1 package; it is based on STM32Cube technology and is easily ported to any STM32F4 microcontroller with an FPU unit.
Information regarding STM32Cube is available on www.st.com at http://www.st.com/stm32cube.
- Real-time sound source localization algorithm (under OpenSoftwareX license) based on two or four digital microphone audio signals
- 180 or 360 degree range
- PCM input
- Adjustable resolution up to 1 degree
- Parameters editable at runtime
- Complete middleware to build applications on top of X-CUBE-MEMSMIC1
- Easy portability across different MCU families, thanks to STM32Cube
- Sample implementation available on the X-NUCLEO-CCA02M1 board mounted on a NUCLEO-F401RE board