VoCon 3200

VoCon® 3200, Nuance’s advanced speech recognition engine, has become the benchmark for high-quality, speaker-independent, continuous speech recognition. The VoCon 3200 engine provides a broad range of unique features, including:

• Highly accurate recognition of natural, conversational input in a broad range of languages
• Ability to handle extremely large tasks such as destination entry for an entire country
• Support for complex dynamic content, such as music titles, in over 25 languages

VoCon 3200’s exceptional recognition accuracy is tuned for noisy automotive environments to deliver a superior user experience and high acceptance rates across a wide variety of in-vehicle systems and applications. Nuance’s R&D organization is committed to the continuous improvement of our speech algorithms to ensure that our solutions meet the highest accuracy standards — even when addressing large and complex tasks.

VoCon 3200 Embedded Development System is a complete development suite that enables developers to add speech recognition functionality to any application. It consists of the VoCon 3200 speech recognition engine, a robust set of development tools, guides, and sample code that allow developers to build a high-quality speech-enabled application with optimum speed and efficiency.

VoCon Toolkit is a set of utilities, meta functions, and ready-made grammars that accelerate development for the most common automotive use cases, including simple voice commands, phone dialing, and navigation destination entry. VoCon Toolkit contains utilities for pre-processing navigation map data so it can be speech enabled with greater speed and ease.

Features

Noise robust front-end
VoCon 3200 is tuned for a wide range of car environments, including mass-market sedans, SUVs, light commercial vehicles, and convertibles

Low CPU & RAM requirements
VoCon 3200 runs on processors and operating systems ranging from ARM 9 upwards. Its scalable architecture minimizes RAM requirements (starting from below 1 Mb) for embedded applications. Low CPU requirements reduce hardware costs for high-quality, yet affordable solutions.

Large vocabulary support (300k+ word list)
VoCon 3200 supports recognition of very large vocabularies for address entry, POI search and music selection by voice.

1-shot destination entry for an entire country
VoCon 3200 allows the user to speak a complete address, including house number, street name, city name and state, in a single utterance for shorter dialogs and more efficient interactions.

Natural Language Understanding (NLU)
VoCon 3200 recognizes naturally spoken utterances. The user is no longer restricted to pre-defined commands when interacting with the speech application, allowing for more conversational interactions.

Music selection with partial title selection
VoCon 3200 recognizes artist and title even if the user speaks only part of the name. This feature significantly enhances the usability of speech-enabled music selection applications because: 1) artist and title names don’t always follow standard language rules; and 2) users don’t always remember the names as they are listed in the database.

Support for multi-lingual input
VoCon 3200 is able to recognize multi-lingual utterances for music and navigation applications. This feature is extremely useful given that Navigation and Music databases typically consist of entries in multiple languages.

Industry’s largest language portfolio supporting over 25 languages
VoCon 3200 meets the needs of global automobile and navigation manufacturers and their customers by supporting a broad language portfolio that includes North American, Western European, Eastern European, and Asian languages.