March 24th, 2009 by Emerging Technologies
If you are accustomed to handling quite a lot of your personal business over the telephone then chances are that you are quite familiar with automated speech recognition. Although this system is wonderfully convenient at times, as well as being rather cost effective, it can be a bother. It would be even more wonderful if all you needed to do was to tell the system exactly what you want. Well, this might be a reality pretty soon, because spoken language understanding (SLU) is a project that is taking on the problem of real-time understanding in the area of advanced telecommunication. This project is called the LUNA project and has a cost of 2.61 million euros. LUNA is headed by Ms Silvia Mosso who is from Loquendo, and Italian speech technology group. They are attempting to make a tool kit for multilingual services that will be able to perform human to computer communication in a way that best suits the average client.
Among the issues that the LUNA project is addressing is ensuring that the system will understand language structure in order to understand what is said. But this is not all. Included in the issues LUNA is tackling are automatic learning, robustness issues for SLU, semantic modeling for the understanding of speech, as well as multilingual probability of SLU components.
The SLU is clearly more advanced than the interactive voice response systems (IVR). With IVR callers have to respond to specific questions by saying specific words or phrases that the system gives as options.
With LUNA that will be a thing of the past. Instead, people will be able to communicate unrestrained with a computer. In the end, thanks to LUNA people will be much more satisfied with the service that they have received.
Ms Mosso explained that they were required to record many unrehearsed conversations between different people as well as between people and computers. The computer’s basic language is called Corpora, and is a collection of all the words and phrases that were recorded. Of course the terms had to be translated in a way that the computers can understand. For example, you might tell the system that your printer is not working properly. It will respond by assisting you to go through a series of options available to you in order to rectify the situation.
Naturally, the developers all agree that once SLU is up and running people will be able to conclude their telephonic business in a far more efficient manner.
Makes the system so agreeable is the fact that it can be useful in any call centre irrespective of what the service the company provides is.
Businesses agree with the researchers that LUNA is extremely practical.
Various academic institutions from around the world are involved in the LUNA project, from France to Italy and from Poland to Japan.
Of course, once the system is fully operational the experts working on the LUNA project are expecting a degree of competition to flare up between the industrial partners.
Whereas LUNA has already devised the most advanced SLU for both Italian and Polish they are still working on the finishing touches. According to Ms Mosso the project should be done by the end of next September.
Read the rest »
March 19th, 2009 by Emerging Technologies
Speech recognition player Nuance has just recently gone public with regards to their acquisition of a percentage of IBM’s speech technology. Having IBM’s source code at their disposal will effectively enable Nuance to make much needed improvements to both its embedded and network-based speech recognition technologies. Inevitably, such an aggressive acquisition, together with the ensuing relationship has many people questioning Nuance’s technology and IBM’s motives.
Furthermore, Nuance has also announced that the source code it has acquired from IBM will dramatically improve its current speech capabilities in both network-based and embedded text-to-speech (TTS) and advanced speech recognition (ASR). Nuance’s intention is to combine the IBM source code with its own over a period of approximately two years in order to enhance the performance of its present speech recognition engine.
While many have speculated that IBM will refrain from competing in this market, IBM on the other hand has made it clear that they’ll continue developing their speech recognition capabilities independently. Interestingly enough, IBM has sold Nuance an earlier release of its code for WebSphere Voice Server middleware and ViaVoice software. According to IBM, the decision to follow through with the transaction was based primarily on the grounds of gaining some return on their investment. However, considering that IBM regularly sells patent licenses, this recent development seems to be nothing out of the ordinary.
Over the course of the last few years, Nuance has been expanding its portfolio in speech rapidly and has made numerous acquisitions. In September 2008 they acquired Philips Speech Recognition Systems (PSRS), followed by SNAPin Software the following month. Nuance has made their intention clear with regards to becoming the leading speech provider in all markets, whether PC-based, network-based or embedded. Furthermore, they can also be seen expanding into complementary areas such as transcription and mobile care.
Although combining IBM’s source code with its own will be challenging, Nuance has long been integrating additional technologies with it’s own so while it may prove challenging to combine the IBM code, they have all the experience they need in order to make it work. Of course, Nuance’s fundamental aim is to develop cutting edge speech technology and purchasing of IBM’s technology has simply highlighted this. It’s hardly surprising that this recent acquisition has resulted in speculation that IBM’s technology is superior to Nuance’s, although if there’s any truth in such speculation, then Nuance’s decision was certainly a wise one.
However, Nuance’s relentless acquisitions have come with a hefty price tag. Regardless of their huge revenue stream of more than $800 million in the 2008 financial year, Nuance has failed to be profitable for a number of years. Having said that, Nuance recently announced that Warburg Pincus, an international private equity company, has just invested $175m in Nuance stock; a move which is seen to be a clear indication of their continued support for Nuance.
Currently, Nuance and IBM will continue competing in the fields of both network and embedded speech technologies. While the additional code being added to its speech technology will inevitably help Nuance to acquire superior software, IBM still has a dominant cash flow and customer base as a result of their other business interests which include IBM Global Services. Admittedly, Nuance is however the most reputable speech solutions provider at present. In fact, they’re the only competitor to offer PC-based, network based and embedded speech technologies. Of course, this carries with it, the advantage of being primarily focused on the speech industry. In all fairness, speech is a very small part of IBM’s business, so Nuance looks set to maintain their lead.
Although there is presently much uncertainty with regards to whether or not there’ll be a partnership between IBM and Nuance in order to develop speech recognition software, such a development would benefit Nuance in particular, as they stand to gain access to IBM’s global customer base as well as their monumental reputation of being one of the world’s largest systems integrators. Furthermore, if IBM does participate in a relationship with Nuance, many will interpret the move as being an indication to the market that they’re backing off.
On the other hand, a partnership could also end up in a conflict of interest with regard to which of the two companies’ would then provide services to speech customers. This may then in turn mean cannibalizing their revenues, a situation which suggests that such a close partnership is unlikely. Furthermore, considering that IBM recently sold some of its speech technology to its largest competitor, many marketing analysts are speculating that IBM many very well end up exiting the market in the nearby future, rather than compete.
Undeniably, Nuance will expand its speech technology and acquire a far stronger market position as a result of this acquisition but for its smaller competitors, there are still opportunities for growth in the network and embedded speech market. Furthermore, they’ll still have the ability to compete with Nuance with regards to competitive pricing. While Nuance is without question the dominant player in North America, it’s presence in Europe is far weaker for network and embedded speech.
Not including IBM, Loquendo and OnMobile in India, which just recently took Telisma over, are now Nuance’s chief competitors in network-based speech. As far as the embedded speech side is concerned, SVOX and iFLYTEK are emerging competitors who like Nuance, have also been investing heavily in their speech technology: SVOX for example recently acquired Siemens’ ASR in January 2009 in order to add ASR to its TTS solutions, while iFLYTEK has just completed ASR development for the Chinese market, thus resulting in them becoming more viable competitors.
In all probability, Nuance’s acquisition of IBM’s speech code will have little, if any, impact on the market, at least in near future. However, over the next couple of years, if Nuance successfully manages to incorporate the acquired IBM code, they will gain superior recognition accuracy in addition to an insight into IBM’s relative position. Utilizing IBM’s technology will certainly assist Nuance to further develop its speech recognition technology faster, with the potentially for restructuring its R&D resources. For the most part, Nuance’s competitors should remain focused on their strengths in both language capabilities and pricing.
Considering to what extent Nuance’s customers stand to benefit, with regards to improvements in technology, Nuance looks set once again to gain an optimum position for competing with IBM. As far as IBM is concerned, most will agree that’s far too soon to predict just how the company will choose proceed, or even if they’ll remain in the speech recognition market.
Read the rest »
March 12th, 2009 by Emerging Technologies
According to American Airlines, their customers will be able to access flight information with the minimum of difficulty thanks to speech recognition technology. This naturally improves its service to its customers, especially those who fly with them often.
Although speech recognition technology does replace the need for human employees it is meant only to increase customer service. However it also means that the airline will be able to save money because the number of its employees has been reduced.
One of the features offered by American Airlines to all its clients is “Remember Me”. When a customer has signed up as part of the airline’s frequent flier program, the system will not only greet them by their own name, but will also recognize their telephone number/s. Each customer is permitted three telephone numbers from which to contact the airline that the system will remember.
American Airlines has said that information such as gate and flight information will even be provided to customers on the day they will be flying without the customer having to actually request the info.
Tellme is the company purchased by Microsoft in 2008 and this is the company that supplies speech recognition technology.
With the global economy in the state it is many companies are looking for ways to cut costs. Because training employees for their call centers represents a large portion of their costs, this technology is beneficial in that way, too, as it saves the airline from having to train staff.
Tellme speech recognition technology has also been sold to other companies besides American Airlines. One other company that has purchased it is Domino’s Pizza.
Of course, as of now this technology is not able to help people to secure a seat on a flight that is already fully booked.
Perhaps it will in the future.
Read the rest »
Tags: 
