cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
martech
Newbie
Newbie
2,016 Views
Registered: ‎10-17-2017

low cost audio processing

I wish to process sounds, in the audio range, and identify up to 10 different sounds. I have never done this before and would like to get some tips on what to read, kits to use and any tutorials relating to this. I want to develop a low cost device for the handicapped. Any help would be appreciated.

 

Thanks

 

martech

0 Kudos
5 Replies
u4223374
Advisor
Advisor
2,007 Views
Registered: ‎04-26-2015

Depending on the processing demands, I'd consider something other than an FPGA. FPGAs are great when you need massive performance, but (a) they do tend to be pretty expensive, both to buy and to use, and (b) audio processing is not normally very processing-intensive. After all, a 15kHz signal (about as high as most adults will be able to hear) gives you over a thousand clock cycles for every single wave on a 150MHz microcontroller (like an STM32F4).

 

With that said, if an FPGA is what you need to get the processing done, then this is definitely the right place to be. My first question would be "are you a student?" - because if you are then Digilent will be pleased to give you a massive discount on some of their kits. In particular, the Digilent Pynq-Z1 is a great board for this sort of project, combining a dual-core ARM CPU with a fairly substantial FPGA. It's normally $229 - but students only pay $65. It's very well supported by Digilent, and it can use all the free Xilinx tools. It's got a built-in microphone, PMOD slots for a few more, and space for an Arduino shield on top.

 

If you're not a student, I'd consider something more like the MYIR Z-Turn, which gets you the same FPGA as on the Pynq-Z1 for about half as much money (compared to the full-priced Pynq). Harder to work with, and it'll need a PCB built to attach it to anything, but at least it makes that PCB design fairly straightforward.

0 Kudos
davewarren
Explorer
Explorer
1,952 Views
Registered: ‎03-13-2014

I also wonder if a FPGA is overkill for decoding sounds. Have a look and XMOS they make devices which can process voice and pick out key words. Also can handle the multiple microphones you will need.

0 Kudos
radhen
Visitor
Visitor
1,085 Views
Registered: ‎05-17-2019

I was wondering how to buy Pynq as a student with USD65 as you said. I am a student and want to experiment it for my projects. Thanks.

Hendar
0 Kudos
kaorichanik
Newbie
Newbie
706 Views
Registered: ‎08-28-2020


@radhen wrote:
I was wondering how to buy Pynq as a student with USD65 as you said. I am a student and want to experiment it for my projects. Thanks.

Hendar

FPGAs are great when you need massive performance, but (a) they do tend to be pretty expensive, both to buy and to use, and (b) audio processing is not normally very processing-intensive. After all, a 15kHz signal (about as high as most adults will be able to hear) gives you over a thousand clock cycles for every single wave on a 150MHz microcontroller (like an STM32F4).

0 Kudos
Steiner_1
Newbie
Newbie
309 Views
Registered: ‎10-03-2020

The correct sound handling will characterize your station's sound, making it stand apart from all the others.

0 Kudos