SVOX, the leading provider of embedded speech solutions, today announced the availability of its next-generation, first truly polyglot Text-to-Speech system, according to CEO Volker Jantzen. Dutch, European French, German, Italian and European Spanish are the first SVOX voices able to speak American English like a native speaker. This means, for example, that the SVOX German voice is able to pronounce an English song name fluently in an otherwise German sentence. Other language combinations are to follow soon.
Marcel Riedi, CTO, explained the difference between the new, truly polyglot SVOX TTS and older approaches aimed at addressing the same need: “In other current systems, foreign sounds that do not exist in the target language are mapped to the closest available sounds. Since most languages have a number of unique sounds, such mapping results in unnatural and sometimes plainly wrong pronunciation of foreign words. In a truly polyglot TTS system, the target language is enriched with additional sounds from other languages, making foreign words sound natural. It is really like when a person learns to speak another language with an exceptional level of proficiency.”
SVOX polyglot TTS was developed to address the strong market demand for a solution that is capable of delivering the expected level of user experience in the ever-growing number of use cases where different languages mix. These include, to name just a few, navigation instructions in countries with more than one language (for example Canada or the US), music track selection or POI reading. For all such applications older TTS systems are often just not good enough.
“It is increasingly evident that the market acceptance for older, non-polyglot generation of TTS systems is falling rapidly, driven by consumers who are less and less willing to tolerate unnatural, funny or incorrect pronunciation of foreign words” noted Martin Reber, COO. “Discussions with customers support our view that for the growing number of fundamental use cases, non-polyglot TTS is simply not good enough.”
Polyglot output is supported by the following SVOX TTS products: SVOX Corporate, SVOX Automotive Pro, SVOX Automotive, SVOX Mobile Pro and SVOX Mobile.
Marcel Riedi, CTO, explained the difference between the new, truly polyglot SVOX TTS and older approaches aimed at addressing the same need: “In other current systems, foreign sounds that do not exist in the target language are mapped to the closest available sounds. Since most languages have a number of unique sounds, such mapping results in unnatural and sometimes plainly wrong pronunciation of foreign words. In a truly polyglot TTS system, the target language is enriched with additional sounds from other languages, making foreign words sound natural. It is really like when a person learns to speak another language with an exceptional level of proficiency.”
SVOX polyglot TTS was developed to address the strong market demand for a solution that is capable of delivering the expected level of user experience in the ever-growing number of use cases where different languages mix. These include, to name just a few, navigation instructions in countries with more than one language (for example Canada or the US), music track selection or POI reading. For all such applications older TTS systems are often just not good enough.
“It is increasingly evident that the market acceptance for older, non-polyglot generation of TTS systems is falling rapidly, driven by consumers who are less and less willing to tolerate unnatural, funny or incorrect pronunciation of foreign words” noted Martin Reber, COO. “Discussions with customers support our view that for the growing number of fundamental use cases, non-polyglot TTS is simply not good enough.”
Polyglot output is supported by the following SVOX TTS products: SVOX Corporate, SVOX Automotive Pro, SVOX Automotive, SVOX Mobile Pro and SVOX Mobile.
