Linguatec develops and markets leading edge language technology solutions for Machine Translation, Automatic Speech Recognition and Speech Synthesis.
Reinhard Busch started Linguatec in 1992 with the potential of artificial intelligence and expert systems in mind. The digital revolution had just begun and he along with a small team of programmers developed software to linguistically analyse large amounts of text. This led to the firm’s cooperation with IBM which back then had a scientific centre in Heidelberg, Germany and was about to develop a new generation of machine translation software which boasted superior performance.
Linguatec licensed this technology and in 2000 they introduced the first translation software for PCs with deep linguistic analysis abilities. The technology became an instant success, and when IBM decided to close all research centres in Europe, Busch had
the opportunity to acquire IBM’s development team following a management buyout (MBO).
Busch’s next step was to develop speech recognition (ASR) as an additional product line for the company. Although it addressed different market needs, it is built on technology similar to that of machine translation with which Linguatec was already familiar. This allowed the firm to further expand their business and therefore to become less dependent on one market.
Text-to-Speech (TTS) is the firm’s latest entry, which they had licensed initially from outside as a component of their MT and ASR solutions. However, Busch goes on to explain this development, saying that when talking to their customers the firm “found that there was significant potential to grow a new market with a standalone speech synthesis solution. Meanwhile we have perfected the technology and our new TTS voices are hardly distinguishable from a real human voice. 40% of traffic operation companies in Germany (busses, subways, trams) use our solution for automating their traffic announcement”. In addition, the firm has important customers in the eLearning sector.
Linguatec is the only company to be awarded the European IT prize three times for their innovative achievements in the field of language technology.
From technology to market
On the marketing side of things Linguatec obtains a detailed understanding of what their customers need and then work out how their solutions can be improved to make their customers more productive. On the technology side, Busch says they “constantly monitor and evaluate the different software technologies relevant for us and then decide which new trends and improvements are worthwhile to be investigated and potentially integrated.”
Ensuring that Linguatec’s visions are executed across all sectors of the company is certainly an ongoing challenge, but by avoiding a purely functional organisation, “hiring people who not only have good technical and functional skills but also share the motivation and pride for high quality language technology products and trying to be a role model for my employees” means that the company is always in a good stead Busch adds.
Linguatec has a dedicated team of computer linguists, language experts and neural computer scientists who share the firm’s mission “to continuously improve leading-edge languagetechnology applications in order to make our customers more productive.”
In regards to hiring staff, “our philosophy is to hire brilliant talent early on (ideally from universities) and train them” Busch explains. “We participate in several research projects (on EU and national level) where we cooperate with the leading universities in our field. This gives us a unique opportunity to identify brilliant people and try to have them work for Linguatec.”
“In terms of the current challenges in the industry, software distribution and pricing is currently changing dramatically” Busch reveals. “The entire retail structure for software will be replaced by new methods of delivery (downloads, apps) and pricing (e.g. subscription models). As for potential opportunities shifting to cloud computing brings new potential for business” he adds. In addition, Linguatec constantly benchmarks their solutions versus what they perceive to be their closest competitors. They also have regular in-depth interviews and meetings with their important customers so Linguatec can better understand their requirements.
In Busch’s previous position as consultant at management consultancy firm Bain & Company, he gained a thorough understanding of various industries. This helps him tremendously to analyse customer needs and to looking to the future to assess market opportunities for new products. Many of Busch’s previous colleagues from Stanford University now work in Silicon Valley which provides him with an invaluable network, for example in spotting new technologies which will be relevant for the company.
To ensure that Linguatec are at the forefront of emerging developments in the industry, they have an international network of professionals to tap into and the firm carefully benchmarks their solutions against competition. In addition, Linguatec has a longstanding customer base in Germany covering many industries, which ranges from large global players to very successful medium-sized firms. In order to integrate technology into their linguistic services “we participate in several research projects which gives us the chance to experiment and test different technologies” Busch reveals. “On the basis of these findings we can make a much better judgement of which components provide proven value to our customers” he adds.
The last word goes to Busch who has an optimistic outlook on the future of Linguatec: “In terms of developing and maintaining our standards, this is very much a continuous process of trial and error. I believe the ultimate benchmark whether we managed to meet our standards or not is to have satisfied customers. “Looking ahead to the future I am very optimistic about the technological breakthrough with deep neural networks. We are currently working on a new generation of language technology products which will lead to significant quality improvements and make our customers even