Hilbert-Huang Transform

Started by woodpecker 6 years ago7 replieslatest reply 6 years ago273 views
Hi Guys,Over the past month, I have been investigating possible methods of signal decomposition as an alternative to the FFT.
In particular, I'm looking at signals containing amplitude and frequency modulated components.

One technique which particularly interests me is the Hilbert-Huang transform, and a quick Google search found this document, which for me was an excellent introduction. The authors give examples of the decomposition of seismic signals, in a simple, non-mathematical manner. R code examples here.

I'm wondering how the HHT compares to using a conventional filter bank. For instance the first intrinsic mode function calculated would seem to produce the lowest frequency component of the signal.

As the document gives coding examples in R, I intend to experiment with decomposition of musical (.WAV files), and will report back if I find anything of note.
In the meantime does anyone have experience of the Hilbert-Huang transform, and if so, would you care to share your experiences ?
[ - ]
Reply by giannipavanNovember 7, 2018

very interesting, I'd like to test HHT with animal sounds and even if maybe not suitable,  I would also very if it can be used to discriminate among animal vocalizations and other types of sounds and noises of not biological origin.

I don't find the link for the complementary materials (source code etc). Do you have it ?


[ - ]
Reply by woodpeckerNovember 7, 2018

Hi Tim & Gianni, 

I have requested a link to the code from the authors.

The R code for Fig 2 (given in the document) seems to run ok, once line 5 is changed to ;

tt <-seq_len(10000)*dt

Also, you may find the hht R package of interest.

[ - ]
Reply by Tim WescottNovember 7, 2018
[ - ]
Reply by UliBruNovember 7, 2018
[ - ]
Reply by Tim WescottNovember 7, 2018

Agh!  You had a link in there, I just didn't see it last time -- I forget that my phone doesn't color-code links well at all.

[ - ]
Reply by Tim WescottNovember 7, 2018

Interesting.  I have a very intense pet peeve about the term "nonlinear signal" -- linearity is simply not a property of a signal, but rather of a system that may have generated that signal.  So that put me off.  But the rest of the paper (for analyzing signals that come from non-linear systems) is interesting.

May be good for separating a signal into chirps or other frequency-varying quasi-periodic components.

[ - ]
Reply by woodpeckerNovember 7, 2018

Ok everyone,

Here is the link to R code examples from the Hilbert-Huang Transform document.

My thanks to the author Danny Bowman for this :

Electronic supplement to The Hilbert-Huang Transform: A High Resolution Spectral Method for Nonlinear and Nonstationary Time Series