Static Helps the Deaf to Hear

OmegaGeek writes: "Jay Rubinstein, a researcher at the University of Iowa, has found a way to improve the signal processing algorithms of cochlear implants (and he's writing in FORTRAN - is this a leading indicator of a FORTRAN revival?). Adding static to the signal actually increases the dynamic hearing range in patients with a cochlear implant."

ummm...

[SuperguyA1]

is this a leading indicator of a FORTRAN revival?

no.

Re:ummm...

[erpbridge]

'nuff said

Re:ummm... - Fortran in scientific computing

[Ilari]

Actually, Fortran still is quite popular in the field of scientific computing. Fortran90/95 and High Performance Fortran that is, definitely NOT Fortran77. F90/95 is actually a rather easy language to program in, it is very similar to Matlab (the leading choise of many scientists for numerical analysis) in many ways, which makes porting from Matlab to Fortran easy. (Many projects start with a rough "first draft" code in Matlab and then move on to more powerful languages as the project advances and computational requirement increase.) Memory management, vector and matrix manipulation is also definitely a lot easier in Fortran than in C.

It still doesn't mean that Fortran is making a comeback. It just fills a particular niche.

Re:ummm... - Fortran in scientific computing

[dpp]

Actually, Fortran still is quite popular in the field of scientific computing. Fortran90/95 and High Performance Fortran that is, definitely NOT Fortran77.

Actually, Fortran77 is still common in astronomy, partly (or mostly?) due to inertia. A lot of code is written in old Fortran, such as the NRAO [nrao.edu] Astronomical Image Processing System (AIPS [nrao.edu]).

During my degree we were taught Fortran90, but during my Ph.D. so much of the old code was Fortran77, and so many of the people you'd work with still used it, that many people ended up writing Fortran77 anyway. Of course, I'm not saying that's a good thing, that's just how it was :-)

It's starting to change, though... the new AIPS++ [nrao.edu] is written in C++, and I haven't written any Fortran for ages.

Re:ummm... - Fortran in scientific computing

[QuantumFTL]

Fortran 77 is certainly not dead, I have a job as an undergraduate at Cornell University, paralellizing a magnetohydrodynamics code in F77 to work on Cornell's Velocity Cluster (it's been mentioned on slashdot before for other research conducted on it). F77 may be a horrid language, which is about as fun as VB to use, but it's still very useful, and it has everything my advisor and his russian astrophysicist collaborators really need - FILE I/O, matrices, and of course we can use MPI, the Message Passing Interface, a well known standard in parallel computing.

You can check out so me of our research at: www.astro.cornell.edu/us-rus/ [cornell.edu].

BTW, I hate F77 but even I have to admit that it's a better choice for them considering they are rather old and do not have time/will to learn anything new programming wise.

excellent work

[Sir Elton John]

As a person with a considerable dedication to music, I am heartened by any progress in the area of hearing restoration. Too often the hearing abled take for granted what surely is the most emotive of the senses. This is a step in the right direction, and a strong step at that.

As for FORTRAN, that doesn't surprise me. FORTRAN has always been the language of choice for low-level signal processing, where the overhead of C libraries makes anything else impractical.

Carry on!

Re:This is equally amazing..

[Anonymous Coward]

It didn't make it on the front page..

Bio-engineering is nifty stuff. From intro undergraduate DSP if the brain doesn't treat each ear independently (kind of like stereo vision) I wonder if the brain is doing any covolution or difference between the signals to increase sensitivity.

Re:This is equally amazing..

[Christopher Thomas]

Bio-engineering is nifty stuff. From intro undergraduate DSP if the brain doesn't treat each ear independently (kind of like stereo vision) I wonder if the brain is doing any covolution or difference between the signals to increase sensitivity.

It definitely does compare signals. This is where a large part of our perception of where a sound is loacated comes from (the rest is from sound interacting with the external part of the ear, which ends up attenuating different frequency bands by different amounts depending on the angle of incidence, if I remember correctly).

RIAA

[Bouncings]

How kind of the RIAA to start encoding CDs so that deaf people listening on computers can hear the music better.

It warms my heart.

Stochastic resonance?

[dpp]

Is this perhaps the same thing as stochastic resonance [hpc.mil] ? I remember reading about it once; it relies on the idea that by adding white noise to a system you can push its behaviour over some detection threshold, and thus convey the signal better, even though you're actually adding noise. Quite interestingly counter-intuitive at first!

From the linked site above:

In fact, there is an optimal amount of noise for doing this: too little noise and the signal doesn't get through, too much noise and the signal gets swamped.

Re:Stochastic resonance?

[mph]

A similar technique, called pre-flashing, was sometimes used for photographic astronomical plates. Photographic emulsions have a nonlinear response, and so you would briefly expose the plate to light (which is a source of noise... we fight light pollution and sky glow all the time) to bring faint sources up to a better part of the response curve.

Ansel Adams also discusses this technique in his books, for improving tonal separation in the shadows.

Re:Stochastic resonance?

[adminispheroid]

Another similar technique is dithering. I guess computer graphics nerds are familiar with this, but it's handy to do to any signal that's about to get digitized -- add about half a bit of noise and you can pick up signals of less than one bit amplitude. Of course with poor signal to noise, but that's better than the no signal to noise you would have had without dithering.

Re:Stochastic resonance?

[Tablizer]

(* Another similar technique is dithering. I guess computer graphics nerds are familiar with this, but it's handy to do to any signal that's about to get digitized -- add about half a bit of noise and you can pick up signals of less than one bit amplitude. Of course with poor signal to noise, but that's better than the no signal to noise you would have had without dithering. *)

It also helps to get rid of the "banding effect" by making the borders of limited-palette boundaries fuzzier, or more random. On the minus side, It can reduce compression size because the noise takes up signal.

Note that you have to do it *before* applying the limited pallete.

Re:Stochastic resonance?

[Farang]

Don't feel bad about that score of Zero: you could be on to something here. Bob Carver used to make a gadget called a "Digital Time Lens" to improve the sound of CD's, and as I recall it not only worked, it added random information to the signal. I was also reminded of the dithering that is a vital part of the digital audio record/playback process. The same, only different. Yeah, I know, score of Zero for me, too. I still say.......

Re:Stochastic resonance?

[WMNelis]

I am one of those people. I have Pink Floyd's Dark Side of the Moon on CD and record. I think it definately sounds better on record. (Sumiko Pro-ject 1.2 record player vs. Sony 79ES CD player) I think it is more likely that the resolution on the CD is too low. But that's just my guess.

On FORTRAN

[mfos.org]

Fortran is still very much used. All the big iron systems use fortran, as many compilers can parralelize (sp?) the code, make it run on multiprocessor machines. It is also very popular with mathematicians, which are its roots.