Since the first Hitchhiker’s Guide to the Galaxy was published in 1979, many people have lamented the fact that the Babel Fish, a “leech-like” creature that feeds on brain waves and can instantly understand anything said in any form of language, doesn’t actually exist.
Luckily, modern technology has taken strides to replicate the usefulness of Douglas Adams’s fictional fish. The translation sector has taken huge steps forward in the realm of speech to speech (STS) translation, with the Global STS Translation Market posting a CAGR of 19.11 percent from 2013-2018.
STS translation technology converts speech into digital signals that are immediately translated into another language in audio format. An STS translation system is used for real-time communication between two or more individuals that speak different languages. This technology is employed in handheld personal devices, smartphones, desktops, laptops, and fixed lines. Computerized translation systems are cost-effective when compared to human translators.
Though the majority of growth in this sector currently comes from the use of this technology in healthcare and government sectors, other segments, including the individual consumer segment, are also expected to deploy these systems and contribute to the market’s growth.
TechNavio analysts have compiled a list of trends spurring growth in this market.
Use of Deep Neural Networks
Current speech recognition systems use the hidden Markov model (HMM) for speech recognition. However, this model is time consuming and occupies a lot of computer memory. Developers are moving toward the deep neural networks model, which has provided better results than HMM on many occasions. Technically, deep neural networks fare better than Gaussian mixture models and are being used for acoustic modeling. Microsoft Corp., a pioneer in this technology, has stated that translation errors can be reduced by almost 30 percent using this technology. While the error rate of HMM stands at one in every four to five words, deep neural networks have an error rate of about one in every seven to eight words.
Demand for Mobile Applications
The demand for STS translation solutions is increasing because of the popularity of mobile devices and consistent network connectivity. Service providers offer applications that allow voice-based access to a cell phone, enabling the use of speech to access mobile device-based or network dependent features, applications, and services.
Emergence of Automated Speech Translation
The Global STS Translation market is witnessing rapid technological changes, and automated speech recognition technology is one of them. Though automated translation has been in the market for several years, automated speech recognition technology has only recently seen major breakthroughs. This technology, which can be integrated with machine translation, is capable of producing real-time STS or speech-to-test translation. It uses a hybrid model that is a combination of rule and statistical methodologies for speech recognition and translation to provide real-time results.
Growing Demand for Cloud-based STS Translation
Until recently, STS translation solutions’ vendors offered handheld devices or software for PDAs. However, the market is currently witnessing a shift toward cloud-based STS applications. Encouraged by the growing popularity of cloud computing, several vendors are developing innovative cloud-based translation applications. Commonly, translation applications are offered through internet-based mobile application stores. These applications are either offered free of cost or at a minimum subscription fee.
