Toshiba's auxiliary speaker cancels noise from main speaker
By Darren Murph 
posted March 19th 2009 9:22PM
posted March 19th 2009 9:22PM
posted March 19th 2009 9:22PM
A look back on popular stories from today in a specific year.
Now that we've thrown 'em off the trail, use the form below to get in touch with the people at Engadget. Please fill in all of the required fields because they're required.
Or even more simply stated, destructive interference ftw
Yow. That's the math end they didn't quite teach us in music school...
Basically, it's impossible to recreate sound without any noise just caused by environmental factors, like 60-cycle hum from your outlet, fluorescent lighting in the room, speaker components vibrating, etc., so what you hear is pretty close to the actual sound, but not exactly what came out of your device, because your playback medium changed it ever so slightly.
This contraption says that it can generate an out-of-phase signal that will only cancel the bad noise...interesting, if it works. Any time I'm doing digital audio work it's tough to clean up noises you don't want without also losing a teeny bit of what you do want. I'm a little skeptical, but okay Toshiba, bring it on.
The link provided speaks of noise caused by the poor construction of the speaker case and build of the speaker components, this isn't about signal noise at all.
Sounds very much like the very old principle of negative feedback, which has been employed in speaker arrays before. It's more a tool for improving distortion measurements than for cancelling noise, but then "noise generated by the main speaker" might be properly classified as distortion, rather than noise.
I've read;
"That said, Josh is planning to conduct research and development in order..."
was like wtf!
ooohhhh gimme gimme gimme!
I was reading the comments and one reader got it right about this tech - normal speakers are not isotropic (Dave) am actually thinking that for this to work would require headphones - which is of course a growth market with a growing number of audiophile products available for it.
Regards
JC
Actually, there aren't "two kinds of noise cancelling headphones". As implied in the name, anything that actually "cancels" noise is going to use destructive phase interference (I.e. Two opposite waves that cancel eachother out.)
What everyone is referring to as "passive noise cancelling headphones" are actually "noise isolating headphones". These devices don't rely on electronics to diminish unwanted outside noise, they do it acoustically.
Anyway, this toshiba thing seems like the wrong direction for a solution to a problem that most consumers don't even care about or understand.
Actually, it seems like the perfect solution to a problem that most consumers don't even care about or understand. The only problem is that most consumers don't care or understand. Then again, I doubt they're overly concerned with what the majority of people care about or understand. It's called the advancement of technology, and I welcome it whether or not most people care or understand. And who really cares about or understands the inner workings of a typical modern gadget anyway? No, what people do care about and understand is whether something works, not necessarily how it works.
Also, is "Kennis" a real name, or did you just throw "Kevin" and "Dennis" in a room together and wait for the perfect name to spring out?
I'm going to take a stab with finite impulse response routine. Take a known input (assumed linear and clean), shove it through the speaker (distortion), transmit the sound waves across the intervening air (frequency dispersion and attenuation), use a mike to gather the distorted, dispersed and attenutated signal, do the fft thing on it, build a correction waveform, do the ifft thing to make a filter. Then the sound that is propagated to the air volume between the first and second speakers will be modified by the second speaker output which is the correction waveform that may be in phase (augmentation) or out of phase (attenuation) for each discrete frequency to correct the non-linearities of the speaker+air and so forth. The beauty is, the correction waveform only has to be generated once and stored for a particular speaker/distance combination. After that, it's just processing power in the digital realm and a routine that once would have taken computer clusters the size of my house can now be done by electronics in a very small consumer device. Ain't technology wonderful? The noise cancellation would just be a serendipitous side-effect of the frequency domain correction. All that is just speculation but sometimes that's where the fun is.
Spot on as I understand it, except that, as other commenters point out, the compensation would have to come from a speaker.. that has distortion and imperfection in output itself, and you could theoretically profile them both and profile the interaction but that would only work in a lab setup with one defined pair of speakers.
Also even if you profile a speaker, I'm sure that cheap speakers even of the same model and factory don't have the exact same profile at all, a bit of glue here or a slightly changed thickness of ABS or MDF there and the profile can change drastically, or even the tension of the wires of the coil due to temperature in the factory could affect a profile, enough to make a compensating speaker actually create an increased reverberating noise.
Most people are missing the point - show me a musical instrument that produces a waveform like either the target or final output trace. It simply doesn't exist.
Basically the technology is all about polishing turd. It has been stated a few times that good quality components eliminate the need of such setup. And laptops etc.? What a waste. Those tweeters have the frequency response of a friggin' telephone, they're wasted by design (and physics - small speakers only produce small sound). Mid-class supermarket computer speakers? Who does expect quality out of them? Most people are happy to hear something out of those. High-end supermarket stuff? Why not just use better components?
I mean, that thing is ridiculous. I have active speakers that cost $300 (meaning they're really cheap) and they're still totally noiseless. Anyone can spend $300 to get decent sound quality - if not, they don't deserve it. Quality costs.
The really bad thing about a phase canceling speaker is that it only works in one direction (the physics laws again, can't change 'em). So you'd better not move out of the sweet spot. In another spot the effect is reverse and the noise gets doubled. Probably the system would be suitable for max. 2 persons at once without severe phase errors in the sound. Funny, eh?
And wiring-based solution of basically the same thing is invented a long time ago. We call it balancing. The principle is basically the same but since it's done in the wire it just kills the noise and nothing else.
atheos,
when you say:
"I have active speakers that cost $300 (meaning they're really cheap) and they're still totally noiseless."
you are confusing noise (hiss or hum heard from loudspeakers) with the impulse response of a loudspeaker (what is shown in the graphs. This mistake is understandable given the incorrect/poor title of the post. Nevertheless, the graphs clearly show the improved impulse response of the device under test.
also, posts which suggest that the use of better components is the preferred solution are incorrect. Even the highest quality drivers have audible distortion (-30 to -50 db range) at certain frequencies. The loudspeaker is the weakest link in the sound reproduction chain and any technology that can make the improvements shown in the posted figure are worth considering.
They acknowledge that the primary speaker doesn't properly follow the waveform that it presented to it. In order to correct the problem, the auxiliary speaker will generate a phase-inverted signal that will cancel out those unwanted acoustic components.
If the primary speaker is unable to accurately create the exact waveform of the wanted signal, how can they expect an auxiliary speaker to accurately produce a precise waveform to cancel out the errors ? The auxiliary speaker will inevitably generate errors of it's own.
Another excellent point, and they say it's to compensate 'cheap ABS thin walled speakers', but to do that you'd need to first use a very expensive lab setup to profile the cheap speaker, and then have a very good expensive speaker to compensate.
Now I guess they could supply a database of profiles of cheap speakers to download, but you'd still need a very good speaker to compensate without throwing its own distortion as you say.
So it's cute but not practical.
Interestingly, their primary speaker is actually an acausal system, as well as the second speaker. They do respond to the original pulse, before the pulse even occured.
And: If it was so easy to predict the speakers noise, why not do some digital filtering instead of adding a second one?
To me, this sounds like crap.
And in the end 98% of the people aren't able to tell the difference between mediocre and good, and the remaining 2% are better served with spending some dosh on good speakers.
I'm probably veering towards being in the 98% category myself.
There are still people reinventing the square wheel!
Anyone remember Dyngroove, the RCA Victor "innovation" from the 1960's? Crappy pickups and arms at the time suffered from inner-groove distortion. Dynagroove employed computers ("highly ingenious electronic brains," said RCA) to generate a reverse distortion to cancel it out.
The result, for a couple of years, were RCA recordings whose last two tracks were unlistenable. After which the "electronic brains" were shelved, and Dynagroove became just a name, with the logo getting smaller and smaller month by month.
Alright, so... If they can make an auxilary speaker capable of perfectly out putting a soundwave that would cancel everything from the mainspeaker that wasn't supposed to be coming out, why not just use the auxilary speaker as the mainspeaker? I'm not really well versed in the science behind this, but isn't the auxilary speaker going to be outputting a soundwave opposite in amplitude to the soundwave from the main speaker except for the parts that you're supposed to hear, and in order for that to work, wouldn't it have to do it perfectly? And, if it could do it perfectly, wouldn't it be easier to just have it output the soundwave you're supposed to hear and have it be the only speaker?