I would like to analysis the audio input source using PD, objective is to know the 'activity' level from the surrounding, such as birds singing, cycling sound, car, people talking, walking etc.
I wonder if 'frequency' is a good property to reflect on this, by definition, it refers to how often the particles of the medium vibrate when a wave passes through the medium. So if more objects make noise/sound, higher the frequency as more particles vibrate through the medium.

So my question is:

1. is there any mic recorder is suitable for doing this? any suggestion?
2. any library can get the frequency result from a sound?

any one can help?

>>2) any library can get the frequency result from a sound?

fiddle~

Hi soon,

Not sure to understand what you mean with "level of activity".
Of course with Pd you can do almost everything but probably this is a case in which
other tools can do better. Maybe this could be useful: tapestrea.
Spectral analysis (analsys in the frequency domain instead that in time domain) will help you
in better discrimination of specific types of sounds.

Mic recorder: there are high quality portable recording systems in the market, also depending on the
amount of money you have. Once I was interested in this:

http://www.edirol.com/index.php?option=com_content&task=view&id=332&Itemid=390

Ciao,

AlbertoZ

Another good one for analyzing your input is "env~". It's an amplitude in db, and it puts it in a range of 0 - 100. Just divide it by 100 and you can get a range of 0 - 1. If you need to analyze the volume of one moment relative to the next, this works well.

007b

In terms of equipment, if you're after subtle, or quiet background or atmospheric sounds, try to use a capacitor condenser mic, as these have the highest sensitivity of all the mic types and therefore can pick up the subtlest sounds, look for low signal to noise ratios in both the mic itself but also the preamp as you will want to crank it up (and therefore make sure you minimize noise)

unfortunately, this sort of equipment also tends to be the most expensive of all the types.

@soon said:

I wonder if 'frequency' is a good property to reflect on this, by definition, it refers to how often the particles of the medium vibrate when a wave passes through the medium. So if more objects make noise/sound, higher the frequency as more particles vibrate through the medium.

frequency refers to how often the particles of the medium vibrate, means how many times in a second. It's nothing about the sound energy.
More objects making sound means more sound energy we can hear. We call
"amplitude". It's about energy - level - volume. So if you want to analysis the sound energy, frequency is not a right way.

mic recorder: Why not use computer,if you want record the sound?
Use Omni condenser microphone for the surround.

there are some externals in rjlib for the rjdj application, that will run in any version of pd too:

http://trac.rjdj.me/wiki/RjLib

The energy is the square of the signal integrated over time. In practice a bank of band pass like an EQ, each with a square, clip and lowpass (leaky integrator) will give you a good "activity" monitor. Use [env~] as suggested above, to get the RMS value of each energy band as a message domain float.