The Speech Market Deserves a Break
The Speech Market Deserves a Break
Speech rec seems poised for growth but it remains to be seen whether recent market consolidations will fuel or dampen its potential.
By Donna Fluss
The speech market can’t seem to get a break. Every time conditions appear right for it to take off something happens to slow its growth. In 1998, after close to 18 years of development, directed speech recognition accuracy finally exceeded 95% on the first attempt, meeting a core quality requirement for call and contact centers and presumably paving the way for its market breakthrough.
But the Web and, in particular, e-service (Internet-based customer service) became hot the same year and the majority of service organizations had to choose between investing in the Web and investing in speech recognition. Even the fact that more than 80% of these early e-service investments didn’t realize their projected benefits didn’t help and investments in speech recognition were pushed aside.
From 1998 to early 2000, Y2K and customer relationship management (CRM) projects ate up a huge percentage of information technology (IT) investment dollars. Just as things were beginning to look up for the speech market, with innovation continuing to deliver improved products with greater benefits and speech interface experts taking over fledgling projects, the Web cooled…but so did the economy.
Speech was again relegated to the bottom of the IT investment list, even though speech solutions would very likely have helped organizations decrease operating expenses while improving service quality. During hard economic times, however, basic survival became the only priority.
Now, just as the economy is starting to see glimmers of recovery, after more than three years of very limited IT spending, SpeechWorks, a pioneer and leader in the speech recognition market and a player that has demonstrated an understanding of customer dynamics, is being acquired by ScanSoft.
Unfortunately, SpeechWorks had very little choice about selling, as they were quickly burning through their cash. The question in the market was not whether SpeechWorks would be acquired, but only who would be the acquirer. SpeechWorks has been one of the visionaries in the speech market place for years and played a major role in defining and building the market but like many market visionaries who came (and went) before, didn’t get to reap the rewards of its hard work. Ironically, SpeechWorks just announced a record quarter with GAAP revenues of $11.3 million for the three months ending June 30, 2003.
No question about it, ScanSoft is a market consolidator – just look at its acquisition history during the past few years. A consolidator is a company that buys other companies for the purpose of aggregating and then distributing its assets. We’ve seen these companies come and go over the years, two recent examples being Divine and Xchange Software, both of which are now gone.
ScanSoft acquired Lernout & Hauspie (L&H), Philips Speech, and now SpeechWorks. Each of these companies was built through acquisition of other speech companies (see the box below). Whether ScanSoft chooses to turn its accumulated speech assets into a viable business for the purpose of reselling the components or to operate it as part of its own business, remains to be seen.
These acquisitions aren’t necessarily negative for the speech market, in the long-term; if ScanSoft can help build one or more profitable speech businesses it may actually be positive. However, ScanSoft is the new kid on the block without a track record in the risk-averse speech market.
ScanSoft claims that its acquisition of SpeechWorks will result in cost savings of $27 million (from head count, office site consolidation and reductions in marketing and administration costs), that it will be neutral to earnings in the first quarter (before amortization and acquisition-related intangibles) and that it will be 5% accretive to earnings in the first full year.
As SpeechWorks already reduced its head count by close to 50% during the last two years and very likely also reduced its marketing budget, it remains to be seen not only where all of the savings are going to come from, but also what the impact of these savings will be on its current product set and SpeechWorks’ customers.
Market consolidation can be positive, but you’ve got to wonder what ScanSoft is going to do with two automated speech recognition (ASR) technologies, two text-to-speech (TTS) products and two or three embedded solutions. Even if ScanSoft is going to maintain all of these different offerings in the short-term, it doesn’t make sense to do so in the long term, as the overhead would be huge and would limit its ability to enhance all of its products.
ScanSoft, like any acquiring company, is going to have to reconcile and consolidate its speech assets, which means that a portion of its customer base is either going to be using a product that won’t be enhanced (and possibly not maintained after 18 to 24 months), or be forced to move to the surviving platform.
Let’s suppose that ScanSoft plans to bring its superior management skills to the SpeechWorks acquisition and that its discipline and management best practices would make the combined companies profitable and the acquisition accretive. It’s been announced, though, that senior SpeechWorks executives will play a major role in the combined company, the same managers who failed to make SpeechWorks profitable, so the theory of an improved leadership strategy appears a bit implausible.
Consolidations are hard in the best of circumstances and very few software mergers are successful. But for the first time, the market momentum for speech is going in the right direction. Perhaps then, the acquisition of SpeechWorks will be accretive for both the market and ScanSoft, bringing positive contributions to this underdeveloped but potentially rich market.