Head-Fi.org › Forums › Equipment Forums › Sound Science › wavelet analysis
New Posts  All Forums:Forum Nav:

wavelet analysis

post #1 of 6
5/10/10 at 8:42pm
Thread Starter 

I thought you may be interested in this topic since I see some posts on Fourier analysis.

 

I used this method to get a signal out of an industrial grinding machine about a year ago. We were looking for signals related to characterizing the size of the crushed material based on acoustic signatures. It worked well enough. lol

 

Compared to a Fourier transform on a time based signal, this method adds the idea of 'when' the signature occurred not just at what frequency.  

 

For example, play a music track and realize that 12.3 kHz was excited, also realize that this happened at 2 minutes and 31 seconds in the track.

 

Not bad.

 

Yup, another engineer geek. :)

 

http://users.rowan.edu/~polikar/WAVELETS/WTtutorial.html

 

http://en.wikipedia.org/wiki/Wavelet

 

http://shadow.eas.gatech.edu/~kcobb/seminar/torrence&compo98.pdf


Edited by ipm - 5/11/10 at 2:10am
Reply
post #2 of 6
5/11/10 at 12:28am

I read several neuroscience imaging papers some time ago concluding that wavelet resampling was superior to fourier resampling in several ways in that field. I wonder if anyone's adapted wavelet-based algorithms to audio resampling?


Edited by b0dhi - 5/11/10 at 5:28am
Reply
post #3 of 6
5/11/10 at 3:58pm
Thread Starter 

It is indeed superior to Fourier since it carries much more information. I will look into this a bit to see if I can find anything. :)

Reply
post #4 of 6
5/11/10 at 6:13pm
Thread Starter 

I am not sure if this is what you are after but here is a mention or two. I hope that helps a bit.

 

http://en.wikipedia.org/wiki/Audio_timescale-pitch_modification

 

http://www.bdti.com/faq/2.htm

 

http://yabb.jriver.com/interact/index.php?topic=29841.10;wap2

 

http://contentdm.lib.byu.edu/ETD/image/etd3247.pdf

 

Reply
post #5 of 6
5/11/10 at 6:24pm

man, i remember taking a course in Wavelet Transforms in my final year of engineering (5 years ago!!!)...was fun albeit the syllabus was a bit abstract.


Edited by sachu - 5/11/10 at 11:24pm
Reply
post #6 of 6
5/14/10 at 10:44am

During the early '90s I worked for a startup that was exploring comercial applications for wavelet based compression/decompression.

 

The thing that was cool about wavelets is even when they are very lossy (think low bitrate mp3s) the way both sound and image degrade is far more natural and less obvious. For example, when you put a highly compressed jpg next to a simlarly wavelet compressed image, the jpg artifacts were much more obvious and visually annoying. Similarly, when you a/b'd an mp3 to lossless compared to a similarly compressed wavelet to lossless, the mp3 was much easier to pick out as different from the lossless.

 

The problem back then is that compression/decompression was much slower or required specialized hardware. Nowadays with the relatibe;y hi bandwidth net connections, cheap disk storage costs, most audiophiles just want the lossless file or feel that high bitrate mp3 is good enough if you have to compress the file.

 

 

Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › wavelet analysis