Vestec's Offerings
Speech has long been considered an excellent medium for data input in a command and control capacity. However, there are several barriers to the widespread adoption and implementation of speech-enabling technology, including the process and the length of time needed to develop applications, and the unnatural quality of user interaction that the technology offers. These barriers create an opportunity for a technology such as ours, which solves these problems.
Based on state-of-the-art soft Soft Computing refers to any computational paradigm which differs from conventional (hard) computing, in that the paradigm is tolerant of imprecision, uncertainty, and partial truth. In effect, the role model for soft computing is the human mind. The guiding principle of soft computing is: Exploit the tolerance for imprecision, uncertainty and partial truth to achieve tractability, robustness and low solution cost. Although meant original for any tool of fuzzy reasoning, the terminology soft computing later included other tools of computational intelligence, including neural networks and evolutionary computing. Soft computing is currently a booming area of research, with potential applications in virtually every area of science and technology.computing techniques, our proposed technologies:
Based on state-of-the-art soft Soft Computing refers to any computational paradigm which differs from conventional (hard) computing, in that the paradigm is tolerant of imprecision, uncertainty, and partial truth. In effect, the role model for soft computing is the human mind. The guiding principle of soft computing is: Exploit the tolerance for imprecision, uncertainty and partial truth to achieve tractability, robustness and low solution cost. Although meant original for any tool of fuzzy reasoning, the terminology soft computing later included other tools of computational intelligence, including neural networks and evolutionary computing. Soft computing is currently a booming area of research, with potential applications in virtually every area of science and technology.computing techniques, our proposed technologies:
- Allow the use of free-ordered, broadly-spoken intentions and indications to command and control a device or system's functionality, doing away with menu- and keyword-driven applications;
- Automate the development of domain-specific natural-languageNatural Language Processing consists of techniques and algorithms for automatic processing of natural language text. Depending on the task, the level of sophistication of an NLP system varies: it can be as simple as word frequency counting or as complex as translating from one language to another or getting the intended meaning of the text. The mainstream research of NLU includes such topics as words' part of speech tagging, sentences' syntactic structure parsing, semantic analysis, etc. NLP is also widely applied in ASR-based systems, mainly in the ASR result processing to turn recognized text into useful instructions for the application and also in dialog management systems to turn recognized text into semantic structures to support the dialog process. understanding engines. Our technology creates rules for discovering concepts from speech input, and can be used with a variety of languages and for any knowledge domain;
- Compensate for the lack of language-training material which would be required to create traditional language-understanding engines;
- Create Natural-Language Understanding (NLU) engines which process continuous speech, even when spoken with different accents, and which are mainly used by enterprises and large businesses.
- Create robust and flexible Automatic Speech Recognition (ASRAutomatic Speech Recognition is a system consisting of computer software (and hardware) that processes digitized voice, recognizes the words spoken and turn them into text, also known as speech-to-text technology. Modern ASR systems applies a set of pattern recognition models (such as hidden Markov models, artificial neural networks) known as acoustic models as the basis for voice to text conversion. A language model is applied on top of acoustic model to provide word combination information for the word search process. An ASR system can be classified as isolated word versus continuous speech systems, speaker-dependent versus speaker-independent systems, etc. For dealing with real-world data, noise removal, speaker normalization, channel normalization techniques are also applied at various stages of the ASR process.) engines and Interactive Voice Response (IVRInteractive Voice Response is a telephony system originally developed for user to hear voice recordings and response to system prompts by pressing dial keys for selecting menu items. A modern computerized IVR system often comes with a much richer set of functions, such as voice mail, database connections. Most notably with the integration of speech recognition technology an IVR system can get user response from their voices and interact with user in a more natural way.) for Small and Mid-sized Businesses' (SMB) voice-based applications.