What's gone so wrong with Microsoft's Speech Server?
A year after Gates promised speech recognition for the masses, the masses have yet to hear anything.
One year after Microsoft launched its Speech Server 2004 with great panache, it has yet to go anywhere.
When Bill Gates personally proclaimed it would bring speech recognition to the masses, Gary Hannah, CEO of speech specialist Pronexus set ambitious goals for his company. He established a separate Speech Server division and "bet a significant part of the farm" on the new Microsoft product.
But when the company closes its books on 31 March, revenue for Pronexus' Speech Server business had come in at about a third of what it budgeted. "It is slower than we'd like," he said. He blamed a marketing effort by Microsoft that was weaker than he expected.
With Speech Server 2004, Microsoft promised to make speech recognition available at lower cost and make it easier to deploy, manage, develop and maintain than competing products. Developers can add speech capabilities to existing Web applications based on Microsoft's ASP framework by adding XML code and SALT (Speech Application Language Tags) technologies (SALT competes with the VoiceXML standard).
But since March last year, Speech Server has had to deal with some setbacks. The product shipped in July, a month later than planned, and JetBlue Airways, one of Microsoft's banner customers, dropped Speech Server for an alternative just as the product launched.
"JetBlue is no longer implementing Speech Server," said Ryan Plant, a software architect at JetBlue. "We have a long-standing relationship with AumTech which has provided hosted speech solutions for us in the past, and essentially they were selected to continue working on speech-related initiatives for JetBlue."
Also, Pronexus and other Microsoft partners discovered that mostly large businesses, rather than the small and medium-size businesses that they say Microsoft initially targeted, are interested in Speech Server. That has forced the partners, both system integrators and resellers, to change their marketing strategy.
Intervoice, another Microsoft partner, has also seen its target market for the speech product shift to large companies. The company realised it was targeting the wrong market and is shifting to larger companies beginning in March, said Michael Segura, director of Microsoft strategy and products at Intervoice.
"I think [Microsoft's] target market is substantially different from what they had originally forecasted," said Segura. He said that larger implementations were proving to be more fertile territory for both his company and Microsoft.
Intervoice is satisfied with the Speech Server business so far. "Adoption through the Intervoice channel is surprising and exceeding our expectations," Segura said.
Pronexus' Hannah is also optimistic about the future, especially since he has seen some changes on Microsoft's side. "The marketing engine is finding the right niche to go after, though it is still struggling. Over the last, probably 60 days, I have seen a difference. There seems to be, at the start of this new calendar year, an interest in listening to the partners," he said.
Microsoft denies that it marketed Speech Server to the wrong segment, sending its partners chasing after the wrong potential customers. "We have been consistent in focusing on both medium-sized and large organisations as the core customer market for Speech Server," said James Mastan, director of marketing for Speech Server at Microsoft.
Customer interest in Speech Server is strong, according to Mastan. He declined to disclose customer numbers, but said that Microsoft has invested a "significant amount of money and resources" in marketing the product. Over 10,000 evaluation copies of Speech Server have been distributed and many deals are set to close soon, Mastan said
"We are actively in the sales process of talking to well over 1,000 customers," he said. "What percentage will actually close varies, but it is going to be a significant number of them."
According to Gartner, at the end of 2004 Microsoft held just 2.7 percent of the market with about 4,000 phone lines, or ports, connected to Speech Server systems. The market is dominated by Nuance and ScanSoft, which held 34.5 percent and 42.6 percent respectively.