LumenVox has improved its Canadian French acoustic models for ASR. We collected additional audio data to improve our phonetic coverage and extended our Canadian French lexicon to include phonetic pronunciation for more commonly used words and names. LumenVox has also added a new Norwegian Female 'Mathilde' TTS voice. The 'Mathilde' TTS voice is also supported in the Speech Tuner.
LumenVox has added additional detailed diagnostic warnings when processing SSML which provide warnings to minor syntax error issues or missing file references that the TTS server automatically skips over. These logs allows the developer to decide whether the default handling by the TTS server is what they expected or if they wanted to fix the issues reported by the warnings.
LumenVox has re-tuned several of its built-in grammars based on customer data. The coverage and accuracy for the time grammar and phone grammars have been significantly improved in Version 13.0. Real world customer data showed that users were using filler words that were not expected by the grammar such as "dash" in a phone number. While the recognizer was often able to skip over the word and provide the right recognition, we found that the confidence score was being affected significantly in such cases. The coverage of the grammar was increased to include these filler words which showed significantly improved confidence scoring and accuracy.
Part of our ongoing commitment is to improve our core speech recognition technology at the base level, which means from time to time we introduce completely new or revised methods of performing some aspects of speech recognition. We?re excited to announce that with version 13.0 a completely new confidence scoring mechanism is being introduced, which offers a significant improvement over our already great methods in prior versions, in terms of real-world results. Our internal testing shows an average relative improvement of over 10% in application performance due to this new enhancement. We also expect to see applications that have a high number of out-of-grammar utterances see an even more significant improvement than this. Be sure to check out this new version and tune in to our upcoming 13.0 release webinar for more details about this significant change.
LumenVox will now support Windows Server 2003, 2008, 2012 (32 & 64-bit), Windows 7, 8 (32 & 64-bit) along with Red Hat & CentOS 5 (32 & 64-bit) and Red Hat & CentOS 6 (64-bit only).
LumenVox is a speech automation software company providing core speech technologies that include the LumenVox Speech Recognizer, Text-to-Speech Engine, Call Progress Analysis, Speech Tuner, and natural language solution support. Based on industry standards, LumenVox Speech Software is certified as one of the most accurate, natural sounding, and reliable solutions in the industry. LumenVox technology provides tools for effectively connecting and communicating with users, increasing user satisfaction, and improving employee productivity. For more information, call +1.858.707.7700 or visit www.lumenvox.com
© 2017 LumenVox, LLC. All rights reserved.