[sdiy] STM32 (or other) audio DSP learning recommendations

[rsdio at audiobanshee.com]

This is great advice that Richie is giving.

I taught a course in Apple’s CoreAudio over the past two years, and I found that I was able to write a program in just a couple of days, complete with graphical debugging displays of waveforms, amplitude meters, and frequency plots. At one point, a local AES meeting featured a speaker who went over the fine details of some useful DSP techniques, and I was able to demonstrate what I’d learned in a new program the next day. Having a computer, with all of the instant debugging and copious display options makes it much easier to figure out what you’re doing, show it to others, and get feedback on problems that might pop up in your math. Embedded development can be much slower because most details are hidden.

That said, Texas Instruments does a surprising amount in their Code Composer Studio tool to make debugging of DSP code easier. When I was working on the TMS320VC5506, CCS allowed me to pick memory buffers with various stages of my signal processing and graph them on the display of the attached computer. The embedded firmware was running on the actual device, not an emulator, but CCS was running the debugger on my Mac. These debug windows included both time domain waveform displays and frequency domain, and you could save a whole configuration of debug windows for later use - saving all the setup time if you have to come back later. I was very impressed, because having these features on the attached computer means that CCS is actually grabbing that memory from the device over the JTAG interface in real time, displaying and converting (some of) the data to the frequency domain. This really sped up my development. It’s still not as fast as working 100% on the computer, but when you have to build custom hardware it’s great to have good tools like CCS.

Brian Willoughby


On Jun 26, 2018, at 10:43 AM, Richie Burnett <rburnett at richieburnett.co.uk> wrote:
> Since you said your aim is to get set up and productive in audio DSP *techniques* you might want to do some experimenting/testing/playing with implementing some common DSP concepts like FIR, IIR, FFT, MinBLEP, etc on a desktop computer platform first.
> 
> I don't want to put you off buying some dsp hardware, but it is often significantly quicker, easier and cheaper to get simple signal processing algorithms running on a desktop PC if that represents a significant milestone or achievement for you.
> 
> Once you get some algorithms working to your satisfaction you can then worry about the additional challenges associated with getting them to run on a dedicated DSP hardware platform, often with limited RAM, ROM, MIPS, power budget, etc.
> 
> Or if you lose the dsp buzz or motivation at the desktop PC stage you have saved yourself the cost of the dsp hardware and its associated Dev tools!
> 
> -Richie, 
> 
> Sent from my Xperia SP on O2
> 
> ---- John Speth wrote ----
> 
> Hi list-
> 
> 
> I’d like to get myself setup and productive in experimenting with audio DSP techniques and I’m looking for recommendations to get started.  My goal is career development (EE and SW eng) that will keep me interested in continuing the project, whatever that may be.  I figured the SDIY crowd can recommend good dev/eval boards with appropriate SW libraries.
> 
> 
> My preference is STM32 based since I know the chip so well but I’m interested in anything that meets these needs:
> 
> 	• Getting started relatively quickly (little to no soldering).
> 	• Excellent SW support (ported libraries, lots of configuration options, high quality code, lots of examples).
> 	• Minimum cost (of course but I don’t want to “cheap out” either).
> 	• Covers signal generation (like a synth) and audio processing (for example, echo/flanging, various filters, vocoder, etc).
>  
> 
> Any recommendations?
> 
> Thanks, John Speth
>