Jetson TK1 Microphone Array - ignore this thread and follow the one in Embedded Systems

Hi guys, I have a work-related project where we want to be able to collect acoustic noises (animals mostly) from a remotely piloted aircraft (small uav) using a Jetson TK1. Some of you may think audio collection from an aircraft is silly but it is very much feasible as it’s been done by others, i just don’t have their documentation so I am doing this from scratch.

Can somebody let me know if the following idea would work or if there is a better way to go about it?

Essentially, I want to deploy an array of approximately 6 microphones onto the belly of the aircraft, connect the 3.5mm jack into an USB Analog to Digital Audio Converter (x6) plug the converter’s USB into a port on a 7-port USB expansion HUB which will then directly plug into the Jetson Tk1 USB slot.

Jetson Tk1 --> 7-port USB 3.0 Expansion Hub --> USB Analog to Digital Audio Converter (3.5mm jack) --> Microphone (powered with a battery).

From what I understand it’s possible to expand the single USB 3.0 port on the Jetson Tk1 with little effort and then program drivers for the microphones. Did I miss anything and is this possible?

By creating an array of microphones this will essentially allow me to perform post processing and beamforming on the audio and filter out unwanted background noises (hopefully). Some of this processing may be done on the board in the future but for now I really just want to capture all collected audio in as clear a signal as possible on the aircraft.

Thanks in advance for taking the time to read and think about this. Please bear in mind I am not an expert in acoustics or programming but I felt with my limited knowledge and the open source community we could make this happen. Hopefully one day we could stop poachers/drug runners from damaging our wildlife habitats(my project).

I think it should be easily possible to connect 6 USB devices to Jetson. There might even be usb “soundcards” that support already multiple microphones so you might need only one USB device with several microphones?

You might not even need any drivers for the microphones. I think there are many standard USB audio devices that work with the same standard Linux kernel driver

And while it would be easy to prototype with 6x USB microphones, that sounds like a lot cables. You may want to plan for the second phase to create your own dedicated mic array. That could be smaller and while USB still is a good choice, you can connect it e.g. to SPI in the extension header. For that you would need to write your own (kernel) drivers.

A small UAV is probably making a lot of noise so filtering that properly out, might be the trickiest part. OTOH, the noise is probably always more or less the same so it might ease the task.

Jetson has a lot of computing power, so doing audio processing in-flight and making flight decisions automatically based on that would be cool :) I don’t think CUDA is designed for audio processing but I don’t see why it couldn’t be used for that too, especially because you will have an array of mics.

The purpose to place multiple sensors like HD-Cameras, temperature sensor, eye tracking sensor, wind sensor on the cap with jetson tk1 board. Such that if any person can go around, wearing the cap and collect the data. The multiple cameras are used to capture the information and data will be stored in the cap.

is this possible on jetson tk1?

please let me know

You’re in the legacy products…you should probably re-post in the Jetson TK1 forum (most Jetson people won’t see this forum):
https://devtalk.nvidia.com/default/board/162/jetson-tk1/

Generally speaking, the JTK1 is probably a good choice for “projects like that”. Depends on your specific needs.