I am interested in the 6 microphone array with SKU: SEN0325.
There are six microphones on this board.
Can an individual microphone be accessed?
Can six audio tracks be simultaneously be recorded where each track represents a microphone?
This so you can use algorithms to separate audio sources?
Example. You are at a gathering full of chattering voices. By applying a statistcal algorithm to all six signals, you can separate the source signals and obtain one voice.
Does this microphone array achieve this with six separate audio tracks?
Reading the description, I can see that the output of this board is designed to be passed onto an a DSP, and that will handle all the complicated maths of working out where a sound is coming from and stuff like that. From that you can infer that the I2S audio stream has 6 channels available.
Here’s an application note for a similar sort of mic I found a while back:
Each microphone has its own chip and access is done via the I2S protocol.
Beamforming is an algorithm to extract the wanted signal from the unwanted signal. The unwanted signal may be noise, it may be a similar signal to the wated signal such as other voices in a crowd.
Ultimatelt you decide which signal to optimize and whoch to reject.
The latter article uses a microphone array similar to the SEN0325 but implemented on the RPi.
Consequently the SEN0325 is “dumb” requiring the user to implement code to perform DSP.
This article gives an implementaion of using SEN0325 on the RPi and a beamforming algorithm.
Here is a practical video on sound source separation.
It ansers the question I originally asked at the start of this page on separation of a particular signal amongst unwanted signals using a microphone array.
The video is in two parts.
The first part demonstrates a person speaking. You hear the speaker and background noise. Then you hear the person speaking more clearly after the source separation algorithm is applied.
The second part demonstrates a crowd of people in the room. Applying the algorithm each person can be heard clearly with the other speakers’ voices attenuated.