Hi all,

I was wondering if anyone could help with this problem.

I'm creating a beat detection patch. The intention is calculate the Peak and RMS of the input signal for further calculation. The Peak abstraction works fine but I am having problems with calculating the RMS. I want to do the calculation in the message domain rather using audio.

So I've been trying to use either [dbtorms] or [env] to achieve this, but the objects seemed to have either a random or zero effect on the input signal. Neither does [lop] low pass the signal which would of been useful.

I am starting to think it might be the method of which I run through the table - using [until] to bang a counter as many times as there are samples in the table.

Is there any other method of RMS calculation that I am missing here? Any information would be much appreciated.

http://www.pdpatchrepo.info/hurleur/BeatDetection.zip

you want to calculate the meanamplitude of a soundfile in an array?
you could do this, as you said, with an until loop or a [kalashnikov] and sum all values in the table and then divide the result by the count of samples.
see the attachment........

http://www.pdpatchrepo.info/hurleur/meanrms.pd

No, sorry if I didn't make it clear.

I don't want to calculate the mean amplitude of the whole sound file.

But rather calculate a running average on each sample in the file. Like how a compressor would calculate RMS. The end result is to have a table that displays the RMS output of the input file.

Also, have you used [klashnikov] instead of [until]? It's seems to have a similar function.

Thanks for your time!

to display an average, you need a certain "window" to do so, you can't display an average every sample. but for this it should work with env~, since env~ outputs an average in a blocksized window (default 64 samples) when i remember correctly. you can also have a look at this pd tutorial on building a compressor, maybe you will find there what you are searching for:
http://www.pd-tutorial.com/english/ch03s09.html
and yes, [kalashnikov] and [uzi] can be used for similar things as [until], have a look at the help files of these objects. the main advantage of kalshnikov is, you can't forget to set a loop end - if you don't set a loop end in an until-loop you will freeze your machine...

Yes I understand the need for a window. I didn't mean calculation on one sample, but that every new input sample should update the running RMS.

The problem with using [env~] is that it I'd have to convert the messages to audio using [sig~] which, if I understand correctly, uses a metro to update. If I am using a series of bangs from [klashnikov] to trigger the Input table then I can't see the calculation being sample accurate if [sig~] updates every so many ms.

I wanted to avoid using audio rate objects as I can extract all the data as messages.

I'm starting think there isn't an object available so I'll have to make the calculation.

Thanks for your help toxonic,

J