White Noise Detector
I’m using this object fiddle~ trying to catch the white noise. I need some help with this method, because I don’t have a clear analysis of the noise, I need to have a trigger only when the white noise is real, not when someone is just talking. Somebody could help me with this?
If what you’re looking for is *really* white noise,
you could always try something simple like
----------begin_max5_patcher---------- 574.3ocyVErbaBCD8L9qfQmotRBI.2a46nSmNxfhix.Rd.4DWmI4auHIfJ2. DZrGO8hPZkX226sKq3kUAfspi7FP32B+dXPvKqBBrlLFB5VG.pXGyKYM1iA3 mXE4uAhbaIOToNnK4Z6l3NqNS5esm67LnQrSxJAQCyB+Q2Q2yz4OHj69YMOW 6NMASVCiBSPlQBseb3UDEVfn193WPd3PH6gAxX60UqLCQKjV4ppJtT26OM+n EMfFMaaIeb5B6rduRpaDmrjEgWCmgZIPCchcrJ0NhgiyM7DjyKnRVkMnf6pE sh5mf1R9ysw5crtnYeKw07IxyjoxyBiBFBtuTwrS5VadLPwkqVTjUsPo1JA2 3jpE9FnVUbMudPSZYEu9ISwbabfvQUp93+LqtJWUppcLCtNiB2jE4OwvTOt8 Ns0IpC6uc2Y9i.SyHwQiNy2sS+8FlZeDisHAk58RsQp3rnAMtOglkgIi.c+z R7roEMeu3bOmX7WJdCAg8870qz9quElP.ylrlp80mnDtWaQtR33YKgIipU3a wG7m30piWut5SHPdYZjqFJaCDswlygcyVVwZWOA578Dn2fdBScwgtlIaDd6b kt6.Ss2ZjAmk4IW.ysNDTJj+8OFXAkw94xQi5PcdeMQ2sxgnAXUvazBISKTR uyXxMg+A6OHJJ3RegoRTrW01hsCCSzB3eARv+ufj4FqP7GgIxYZ4GhIZaSbT RRRVpo3v8uSYoqocqtTHSVhLFeakQ5RvD4BvT6hWW8a.toRiTB -----------end_max5_patcher-----------
Gregory? Just curious… wouldn’t you want to measure phase coherence ? A sine sweep up the spectrum takes that meter from empty to full, none of that’s noise. Is this more of a pragmatic approach? I’ve never seen it before.
any fft-based analyzer with something like 4-5 frequency sections,
no matter at which exact positions, should work fine for finding
"white noise", even fffb~ maybe.
with tristan´s it is easier of course.
You might want to give a try to zsa.descriptors. zsa.spread~ or zsa.slope~ might do the trick.
|Matthew Aidekman wrote on Wed, 09 September 2009 01:45|
|A sine sweep up the spectrum takes that meter from empty to full, none of that’s noise. Is this more of a pragmatic approach?|
I bet it is. any fixed high frequency will produce enough zero crossings to count. In the scenario David describes it seems perfect, as you usually won’t have just high frequencies on a sustained sound. The normal frequency distribution is characterized by a majority of relatively lower frequencies, and with speech, a lot of pauses…