Hello pd community!

I am actually doing a project involving a raspberry pi and pure-data, the aim is to create a sound level meter that calculate the equivalent level of the sound in dB (A-weighted). Basically, it is just a stockage of the level every seconds and ponderate on a 15mn basis. [if you don't know what is it : http://www.engineeringtoolbox.com/equivalent-sound-level-d_721.html]. The input~ will be provided via the wolfson audio card and the calculation on pure-data.

I started to create my patch but i struggle a lot, i would like to know if anyone would know existing patches that are used for that application : sound level meter, dbA filter, level stockage, any acoustics measurements,....

Anything could help

Thanks in advance!

Hi @buglover,

I am also using a RPI with the wolfson audio card. I have successfully compiled a RT-kernel and using PD with low latency settings. Really nice combo

What kinds of problems are you facing exactly? I am not sure if your question is about the maths or the patch?

Cheers~

Hi @EEight,

I studied the maths about signal theory, DSP, buffers, filters, ... I worked on a texas instrument card and codecomposer. Even if it was a few years back, I still have some notions of that. My troubles comes when i tried to applied that on pd, i am not even working on the pi already.

What i tried to do is : Input~ => dBA weighting filter => SPL + calibration => Leq 1 sec => storage => Leq 15mn
I was wondering is someone has ever done something similar....

I am currently at the first step with the dBa filter which is a IIR filter. I tried with fexpr~but without success (dBa_IIR2.pd. The next idea is to use FIR filter so i need to convert the coefficients via prony's method.

I thought using that distribution for the pi [http://forum.pdpatchrepo.info/topic/8476/raspi-wolfsonaudiocard-and-pd], is it better with a RT-kernel ? (sorry for the newbie question)

Regards

Hey buglover, that's a seriously cool project! Building an actual calibrated SLM in Pd is pretty deep stuff.
For the dB(A) filter, I'd definitely recommend checking if any of the external libraries you're using (like iemlib) have a pre-cooked filter for the A-weighting curve. Trying to implement that IIR from scratch can be brutal, as you found with fexpr~!
And yes, absolutely go for the RT-kernel on the Pi. For any real-time audio and especially for continuous logging like an SLM, having that low-latency stability makes a huge difference and will save you endless headaches down the road.
You might find this free tool helpful. https://noisemeter.co/

@HannaGen said:

a pre-cooked filter for the A-weighting curve

not sure if precise enough (although the error curves look really good!) - but here is a nice source for biquad coefficients for an A-weighting filter at 44.1 and 48 kHz:
https://jenshee.dk/signalprocessing/aweighting.pdf (from https://jenshee.dk/signalprocessing/signalprocessing.html)

In Pd, at 48kHz, this translates to:

[biquad~ 1.34731 -0.349058 0.965251 -1.3473 0.382051]
|
[biquad~ 1.89387 -0.89516 0.94697 -1.89394 0.94697]
|
[biquad~ 1.34731 -0.349058 0.646665 -0.383622 -0.263043]