few days ago i heard about this "spactral analysis of spactral analysis" wich is called cepstrum. i was searching on the net, but there are really few sources about this, and no software at all!
i'm really interested in this stuff, so if anyone can help me with some interactive video, tutorial or a good URL, i'd be more that thankful.
Cepstrum, Quefrency, Gamnitude, etc... WTF???
-
- KVRist
- 69 posts since 16 Mar, 2003 from Croatia
-
- KVRist
- Topic Starter
- 69 posts since 16 Mar, 2003 from Croatia
i found some little info, but i still don't get it - does this thing fit into DPS/synthesis ???
anybody?
anybody?
-
- KVRAF
- 12235 posts since 18 Aug, 2003
It's just an analysis technique isn't it? Used to figure out the periodicity of spectra.
Any software I've seen with it implemented as an analysis option is both a)designed for academic/engineering purposes and b)costly. The original paper also seems to be more of a mathematical parlour trick than anything earth-shatteringly useful. Also it's part of the Fourier Transform rather than Fast FT, so likely it's computationally impractical for muscial purposes.
I googled about, seemed to pull up quite a bit of info.
Any software I've seen with it implemented as an analysis option is both a)designed for academic/engineering purposes and b)costly. The original paper also seems to be more of a mathematical parlour trick than anything earth-shatteringly useful. Also it's part of the Fourier Transform rather than Fast FT, so likely it's computationally impractical for muscial purposes.
I googled about, seemed to pull up quite a bit of info.
-
- Skunk Mod
- 21249 posts since 10 Jun, 2004 from Pony Pasture
Sounds as if it's either electronic dyslexia or a close relative of one of the predominant radio broadcasting methods: Momplitude Adulation, in which songs of praise to mothers are all you transmit.
-
- KVRist
- 190 posts since 28 Nov, 2003
You can truncate the cepstrum and then work back to the time domain to get a minimum-phase version of a sound (the cepstrum (log(dft(x))) allows you to separate the minimum phase transfer function a(z) from the allpass b(z) (because log(a*b) = log(a) + log(b)) ). Should be a link around somewhere... http://www.tsp.ece.mcgill.ca/Kabal/pape ... uC1995.pdf . Might be tricky to get good results for real-time deconvolution, though 
-
- KVRAF
- 4908 posts since 10 Aug, 2004 from Colorado Springs
Cepstrum analysis is not just a parlour trick as someone suggested.
In non-musical noise analysis, many times the annoyance of a sound that has tonal aspects to it, is determined by spacing of tonal components in the frequency domain. If these are evenly spaced apart from each other (as they often are), then cepstrum analysis allows you to measure the strength of that tendency. Often, an amplitude modulation results that has the modulation frequency of the spacing between the tones. Most folks in the automotive noise analysis world (of which I live in when I'm dreaming about playing my guitar), nowadays use modulation analysis instead of Cepstrum. But for many years, Cepstrum analysis was the only game in town.
If you are familiar at all with the psychoacoustic measures of Roughness and Fluctuation Strength, both of these two look at the modulation spectrum and apply weighting curves to them that were empirically determined.
I don't know that Cepstrum would be all that useful in a music/audio processing mindset.
-Scott
In non-musical noise analysis, many times the annoyance of a sound that has tonal aspects to it, is determined by spacing of tonal components in the frequency domain. If these are evenly spaced apart from each other (as they often are), then cepstrum analysis allows you to measure the strength of that tendency. Often, an amplitude modulation results that has the modulation frequency of the spacing between the tones. Most folks in the automotive noise analysis world (of which I live in when I'm dreaming about playing my guitar), nowadays use modulation analysis instead of Cepstrum. But for many years, Cepstrum analysis was the only game in town.
If you are familiar at all with the psychoacoustic measures of Roughness and Fluctuation Strength, both of these two look at the modulation spectrum and apply weighting curves to them that were empirically determined.
I don't know that Cepstrum would be all that useful in a music/audio processing mindset.
-Scott
Last edited by rockstar_not on Tue Sep 27, 2005 12:29 pm, edited 1 time in total.
-
- KVRist
- Topic Starter
- 69 posts since 16 Mar, 2003 from Croatia
thanks guys!
i just wanted to be informed, and during the lack of resources i asked on my favorite board
i just wanted to be informed, and during the lack of resources i asked on my favorite board