- Faster G2P (grapheme-to-phoneme”) calculation reduces the enrollment time for newly plugged-in music libraries or large phone books;
- New enhanced acoustical models (HMMs) for German and US English allow for best in class recognition results and one-shot destination entry;
- Support for Word Hypothesis Graph enables more intelligent interpretation of recognition results, e.g. by the SVOX Dialog solution;
- Introduction of optimized algorithms results in shorter response times and lower CPU consumption.
The new release is immediately available to SVOX automotive and mobile customers as a part of SVOX Automotive ASR and SVOX Mobile ASR solutions in various configurations.
“This latest improvement of SVOX core Speech Recognition technology allows our automotive and mobile customers to achieve even higher recognition rates in even more difficult noise conditions,” said SVOX CEO Martin Reber. “We are now in this exciting moment where advances in the core technology and adoption rates fuel each other to produce a virtuous circle. It is especially the case in the mobile devices market where we currently see not only higher ASR and TTS penetration rates, but also significant improvements of user experience because of speech.”
SVOX Automotive ASR and SVOX Mobile ASR Speech Recognition solutions are highly portable and available in the following languages: Canadian French, Czech, Danish, Dutch, Finnish, French, German, Greek, Italian, Japanese, Korean (2010), Mandarin Chinese, North American Spanish, Portuguese, Russian (2010), Spanish, Swedish (2010), UK English, US English.