|
|
|
|
Our expertise in speech interfaces spans the entire range from speech algorithm development to speech applications, for both the desktop and the telephone. We worked on the first commercial continuous speech recognition system at Verbex, designing algorithms in software and in special purpose integrated circuits for high-intensity computation.
We've created a multi-application conversational speech system, including dialog management, natural language generation and inference engine-based understanding. We've created optimized grammar compilation tools to generate large finite state grammars automatically using a high level grammar specification language (using a domain lexicon and syntax templates).
We've created speech applications at customer sites and tuned them for production use. We are familiar with speech API's and tools for ViaVoice, Dragon and Philips on the desktop, and IBM, Nuance and SpeechWorks over the telephone.
We've also assisted with new product planning for speech recognition vendors, and performed speech product evaluation for venture groups.
We've used speech and pen, as well as standard GUI interfaces, to create intuitive, easy to use applications. We've developed multi-platform (Unix, Windows, Apple) applications utilizing platform independent GUI APIs.
We've created complex speech and image processing applications, using both standard CPUs and DSPs. One example is an in-house tool for displaying and analyzing speech spectrograms in real time using Intel's NSP libraries.
We've used object-oriented design and pattern-based architectures to develop complex applications that are easy to understand and maintain. Clear design documentation (including UML) and thorough unit testing are our hallmarks.
We've developed object-oriented Product Data Management and Knowledge-based Project Management systems, in n-tier architectures using Oracle and SQL Server backends. We also developed a CASE tool for C++ object-oriented development.
We've used artificial intelligence techniques in a wide variety of applications, to make software smarter and easier to use. We've developed an inference engine employing truth maintenance, forward and backward inference, and an expressive domain model-based representation for knowledge. We've used this inference engine to develop numerous expert systems, including a process control expert system for designing distillation control systems.
We've used other commercially available tools (such as NeuroSolution by NeuroDimension) for neural network, fuzzy logic and genetic algorithm development.
We've explored complex knowledge representation in the email domain using Cyc, a large common sense reasoning system developed by Cycorp.
We've internationalized existing English-only applications, as well as designed new ones with internationalization considered from day 1. We have localized software for the German and Japanese markets. We've also designed speech applications for the global marketplace, dealing particularly with grammar translation issues.
We've implemented large, complex real-time embedded systems. Some examples include a multiprocessor, multiconfiguration intelligent chemical analysis and separation system, as well as several generations of special-purpose speech recognition devices. We have used real-time operating systems in these projects, and well as doing extensive programming at the assembly language and microcode levels.
We've developed highly optimizing compilers for array and parallel computers. These compilers have used AI technologies like constraint propagation, symbolic algebraic manipulation and complex planning and scheduling techniques to make good use of intricate hardware architectures.
We are experienced with creating web applications using HTML, DHTML, ActiveX, Java, and JavaScript, as well as providing web access to server applications. We can also create voice web applications using VoiceXML.
We are experienced with C/C++, Smalltalk, Java, JavaScript, HTML, XML, SQL, Visual Basic, LISP, Pascal and Fortran; assembly language for 680x0 / 80x86 / 320x0 families, VAX, PDP11, Cray2 and IBM370, as well as microcode for custom architectures.
We are comfortable using Win2K/9x/NT/WinCE, Unix/Linux, Apple, DOS, OS/2 and VAX/VMS operating systems.
We are experienced with cross-platform and native Windows development (using COM/MFC/Visual Studio).
For database development, we have used Oracle and SQL Server.
For speech development, we are familiar with SAPI 4&5, SMAPI and VoiceXML. We are familiar with a variety of expert system shells, fuzzy logic, neural network and genetic algorithm development.
|
|
Send mail to
webmaster@reasonix.com with
questions or comments about this web site.
|
