JetBlue dials in to Microsoft’s speech-recognition technology

When you think of speech recognition (or voice recognition), you might think of the simple speech-to-text processing tools such as Dragon NaturallySpeaking, which transforms spoken words into text within your document-processing software. You might also think of some of the more common applications we are becoming comfortable with on a daily basis such as voice dialing (like when you’re driving along and say “Call home” and your command is obeyed) or voice-entry scenarios (where you are asked to say “yes” or” no” or to provide a number verbally).

But the applications for voice or speech recognition are becoming increasingly broad — and cutting edge. More importantly for those who care about the bottom line, they are becoming financially justifiable. And who is at the center of this push toward speech-recognition solutions? Microsoft. Who else?

A recent move by JetBlue is allowing the airline’s passengers to retrieve flight information with a voice-activated interface when they call in. It’s a simple scenario that most of us are familiar with: You call a toll-free number, speak the flight numbers, arrival and departure cities, and flight times, and you are provided with the information you need.

Here is the problem: If the system doesn’t work well (and we have all dealt with voice-activated solutions that drive us to cry out, “For the love of God, give me an operator! A human!”), it only aggravates passengers and the call centers are still bogged down. If it does work, well, passengers are happy and agents can utilize their time to focus on more important matters with JetBlue customers. But it also has to be affordable. Well, the headlines from a recent Webinar regarding JetBlue’s phone changes show the affordability side: “How JetBlue is Replacing their $1,000 Speech Recognition Licenses for $15“.

JetBlue chose to go with Aumtech, a provider of IVR (integrated voice response) and instant notification solutions for automated delivery. Aumtech paired its Aumtech Responsive Communications (ARC) telephony suite with the Royal Philips Electronics SpeechPearl 2000 speech recognition engine to create the system JetBlue is using.

So where does Microsoft come into play on all of this? Well, there are usually steep costs involved with speech recognition. Aumtech, however, collaborated with Microsoft to leverage the speech recognition services built into Office and Vista, resulting in a lower-cost solution. By lower cost, I mean 75 percent lower than competitively placed solutions, according to some accounts.

Here are some of the numbers: When JetBlue wanted to increase its existing Philips/Nuance ASR application license (which had an 88 percent accuracy rate and 60 percent call completion rate), the airline was looking at a price tag of $175,000. Aumtech said it could do it better and for less, $47,500. JetBlue says the system now has an accuracy rate of 90 percent and call completions are at 82 percent with the capability to handle three times the call volume than the previous system.

According to TMC blogger Brendan Read, who just returned from SpeechTEK in New York City, “The Aumtech/Microsoft/JetBlue story is shaking up vendors in that space because of the pricing model; also anytime Microsoft shows a strong interest in a market there is concern by smaller firms of what could happen to them. At the same time there is interest by those suppliers who make applications using the engines in Microsoft’s low-priced and rugged speech solution because it could help them cut their costs and become more competitive.”

As I mentioned, Microsoft’s speech-recognition technology already exists within your Windows systems, and Vista really shows it off in the Ease of Access Center. You can dictate documents and e-mails. You can use voice commands rather than keyboard strokes to open and work with applications. You can control the entire OS really and do pretty much whatever you want through your voice. The Windows Speech Recognition tools in Vista can adjust to your style and vocabulary too.

Personally, I’ve worked with the new speech-recognition tools and I thought they were amazing. Easy to use and, best of all, it understood my New York accent without a problem. The only thing I found frustrating was remembering the right commands to have the OS do what I need. But I imagine that, as with texting, I just needed a little more practice. But one of the cool things that Microsoft did was not just provide speech recognition, but also the ability for developers to speech enable applications through a new WinFX namespace, System.Speech.

Microsoft has developed its own Speech Server, designed to work with IVR applications that utilize telephones, cell phones, mobile or tablet devices, desktop communicator tools, Web applications and so forth.

One area where Microsoft has really decided to eat its own dogfood is with its Exchange 2007 Unified Messaging Server. This features an automated attendant, allowing people call in IVR and request their mail, calendar information, voicemail and so forth. The auto attendant can be configured to handle different scenarios and languages.

So, where else might we see speech- or voice-recognition-oriented software in our world? Logically, one is the military, and testing has been ongoing for fighter jets. Another area on the other end of the spectrum is health care; medical transcriptions, for starters. Moreover, people with disabilities benefit from advancements made in speech recognition — and we can see this up close in Vista with the speech-activated capabilities already integrated into the OS.

So, with companies such as Aumtech working with trusted technologies like Microsoft Speech Server, we can hope to see more affordable solutions continue to be developed. But we can also expect to see auto attendant systems that don’t irritate and annoy callers into that aggravated frame of mind where they just push 0 on their phones and scream “I said, ‘Yes! Yes! Give me an operator! @##@!'”