Low-level white noise, better transcription accuracy?

 

Yes.. you read the title correctly!

In an article posted 11/27/2007,  HealthImaging.com has noted a study at the University of Maryland Medical Center - "found that the introduction of white noise at certain levels as part of the acoustic background increased accuracy of speech recognition systems’ transcription capabilities", said Joseph Zwemmer, MD, who presented the results of the research.


This is actually pretty fascinating stuff. If white noise can be used to improve accuracy, then it's more than likely another step has been made, albeit maybe inadvertently, in the proverbial quest for the "Holy Grail" of speech recognition; to-wit software-based noise canceling.

The original article can be read here.

Hopefully, we'll be able to get some specific data soon from the University of Maryland as to what exactly comprised the white noise, what frequency ranges it included, details as to the sound itself, durations of specific frequency ranges, etc.

This has some very interesting possibilities..
Check back from time to time to see if the Medical Center has shared any data,
on just what type of sound the white noise was comprised of..

Cleaning the Loo and recognizing speech..

 

Who would have imagined?

A group of deleopers for West Nippon Expressway Company Limited (NEXCO), one of the main operators of expressways in Japan has come up wht a "bug" that cleans the bathroom - and recognizes speech!

It's makers say it speaks politely, and is being designed for cleaning public restrooms at highway rest areas. It even comes with "access to current traffic information in case someone has the urge to spark up a conversation",,
How cool!

We posted an article about US market validation for speech recognition recently; it seems there's speech recognition enabled devices springing up across the globe, too.

Yes, it's a cliche musing but;
We wonder what's next..?
.

Google -and- Microsoft; Betting on speech-based searches!

 

An article from MySpaceNews has another reference to..
"The development of increasingly precise speech recognition technology begs the question of whether Google and Microsoft will one day use it to "listen" to users, in order to serve them relevant advertising".

The article talks about how both Google, and the leader in speech recognition, Microsoft, are both building speech data banks.
Why?

Both internet giants are.. "using their free directory services to build vast, digital Petri dishes loaded with samples of users' speech patterns. Those patterns are in turn are being fed into the companies' myriad server computers, where they are analyzed in an effort to make their respective speech recognition technologies progressively smarter."

And again - why might that be, according to rumors?
"accurately understood search commands could be paired with equally relevant, and valuable, advertisements."

Nonetheless:
Rgardless of any presumed dark motives, the data these two highly evolved industry giants will quickly gather and analyze will undoubtedly become a giant leap forward in the holy grail of speech recognition to everyone - The proverbial Holy Grail of speech recognition for over a decade:
       "Speaker independent speech recognition".

What a terrific development. It's more than possible a year from now all the wireless devices across the world will be interacting with websites in very new, very exciting ways!
.

Skype + Speech Recognition, powered by Google?

 

Both Rich Tehrani at TMCNet and Jemima Kiss (PDA: The Digital Content Blog) have posted articles in the last few days about a circulating rumor..

Maybe Google's looking at a Skype acquisition?

This interests us because Rich has some cool conjecture(s)..

Rich wonders..

"You don’t think there are 100 propeller heads (said respectfully – I am a fellow propeller head) in the Googleplex working on speech recognition right this moment so if they acquire Skype they can immediately run contextual ads based on what you are talking about?"

and we've known Rich for over a decade... Rich isn't often wrong!

Of course, if this happens we hope this means to Skype users everywhwere (on wireless headsets, of course :-) that they'll be generating speech recognition, without even trying!

Dragon's in the air -- Is the market finally validated?

 

For the last several mornings, we've noticed a rather well done commercial for Dragon Naturally Speaking's speech recognition software on major news networks, every morning in the AM.

The commercial shows an executive dictating a document, a lab tech finishing and sending an email and the executive's son starting, dictating & sending an IM. The commercial finishes up with statements such as "Windows Compatible"; "Talking is 3 times faster than typing"; "Up to 160 words a minute"; "Up to 99% accurate".

(It's also interesting that from a marketing viewpoint, this commercial shouts "simplicity" in every respect!).

In an interview with CNet News, when asked about why more light isn't shined on Windows' Vista's amazing speech recognition - Bill Gates talked about speech recognition migrating from being what many call an "obscure feature", into something more prominent:
"The speech recognition comes into it--all these things about natural interface are coming to the fore, and they are probably the thing that's most underestimated right now about the digital revolution" and "When voice recognition is used in the right way--let's say you're in the car and you want to pick somebody to call--that's improved very dramatically.."

Yes, it has! For those of us who have been using speech recognition for a decade or so, the differences are like comparing a wood stove to a microwave oven.

Although the website listed in the TV ad is a cheesy re-direct to a "Buy Right Now" ecommerce order page in Digital River, with absolutely no good copy to entice the curious should-I-buy-this seeker, nonetheless..

It seems to us, when a large, smart company like Nuance will plunk down the dollars necessary to have it's Dragon product ride the airwaves day after day, that's a pretty darn good indicator this part of the digital revolution is "ready for prime time!"

.

Macros are coming to Vista!

 

.
Cool!

Rob Chambers, has just announced that Microsoft is releasing a Macro Tool, for Windows Vista in the near future.. a much anticipated, highly requested addition to the superb recognition that only Vista can offer.

You can read about it here.

Even cooler, the tool can be customized with XML, which can be a real snap.

Our hats are off, again, to the venerable Speech Components Group!

.

Wireless Speech Recognition in the ER

 

An excellent speech recognition blog, (appropriately titled "Speech Recognition" :-) has posted a really interesting article about an intriguing remote-based wireless speech recognition system for Emergency Rooms, and also links to a terrific case study on the Crescendo site.

The case study's diagram outlines a system that uses a remote central voice & data server_+_Speech Recognition (SpeechMagic) Server that is accessed via Pocket PC's, and allows ER physicians to "walk and talk" while maintaining access and the physicians are provided with up-to-date patient data on their PDAs - any place, any time.. driven via speech commands.

It's fascinating reading, really worth checking out.
.

The "Light Link" for speech communications

 

.
Laser communications systems are wireless connections that work similarly to fiber optic links, except the beam is transmitted through free space.

The U.S. government's new Lasercom Interoperability Standard has been tested by Boeing, and demonstrated to the Military Satellite Communications (MILSATCOM) user community that it can rely on a 10 to 40 gigabit per second TSAT backbone between satellites. Just imagine that kind of bandwidth ..

TSAT is designed to provide the secure, long-haul, high capacity backbone and full Internet-like functionality for U.S. forces to ensure that they have information superiority in any situation throughout the world.

The Japanese Experiment Module "Kibo" on the International Space Station, which can now attach to & interface with the ISS, will carry the 1st laser communications experiments based on TSAT technology. This experiment will demonstrate large-capacity, high-data-rate (2.5 Gbps) two-way communications, between the ISS and ground stations.

As to what that means: Real-time voice communication, among other things.
- Talk about "lightning fast" wireless remote speech recognition ..
Across how many miles .. ??

_________
Update
____

One of our readers left a comment that asked:
" ..surely the latency is the same as laser and radiowave both travel the distance at the same speed? "

We posed this question to our leading wireless audio engineer.

Here's what our engineer replied with:

"Light travels at 3e8 meters per second or 186282 miles per second.
The ISS orbits at (approx) 220 miles from Earth:
220/186282 = latency @ 0.0011 seconds each way. Light and radio travel at about the same speed thru space. The latency from the Moon to the Earth would be 1.3 seconds, each way, for example. The data rate does not really come into play unless you are compressing the data so much that your codec adds up to 100ms of extra delay.


Ultra Wideband at 60GHz can do a few Gbps - but only has a range of 20 feet or so. It really comes down to available bandwidth and transmit power, and power loss thru the medium you are transmitting through.."

The NASA sources we put this question to appear to base their position on throughput.

They note that RF's maximum throughput (presently) is around 100mbps, per 802.11n, where laser communications offer bandwidth to 2.56 terrabits per second (presently).

.

Bringing Harmony to Space

 

At 1:01:15 PM ET today, space shuttle Discovery landed at NASA's Kennedy Space Center in Florida.

It was a fairly quiet mission media-wise, but it was one of the most important Space Shuttle missions in the history of the program.

Mission STS 120 carried something called the "Harmony Node 2" about which NASA says:

"Think of the next component set for delivery to the International Space Station as an international crossroads in space. That's the major function of the Italian-built U.S. module that will be ferried to the station aboard space shuttle Discovery during mission STS-120."

Harmony provides a passageway between three station science experiment facilities: the U.S. Destiny Laboratory, the Kibo Japanese Experiment Module, and the European Columbus Laboratory.

The international crossroads the Harmony 2 creates brings up to space something that, in the hardware world is a notable occasion. The European Columbus Labratory will now comfortably attach and exchange with the ISS; and NASA Engineers who tested the communications equipment of the Columbus said this, when finished:

“This is the best piece of hardware we have ever tested”.

 … Wow! … 

"Comparing Apples to Gold Nuggets"

 

To quote Rob Chambers -

"The issue, as this author points out, is that people believe that the speech recognition in Apple's OS X can do far more than it actually can do. But, effectively, it can only do some limited command and control scenarios. Vista can do so much more. "

Comparing the speech recognition in Vista with speech recognition built into the Apple OS X is like comparing Apples to..
        Well - the title says it all.

.