BCL Technologies offers a new way to communicate with desktop computers and handheld devices using Spoken Language (SL) Interfaces. In general, current speech recognition systems force users to use an extremely strict command set. These systems cannot handle slight deviations from pre-defined sets of phrases, and therefore do not allow humans to communicate naturally. BCL's Natural Language (NL) capabilities offer the freedom for users to communicate with voice-enabled handheld devices and desktop computers using totally natural SL with any number of variations in which a specific command or query may be stated.

BCL hold several patents in this field. Its research is published widely in refereed journals and peer-reviewed conferences.

BCL has a suite of technologies in the area of NL/SL. This section provides a brief summary of the breath of this technology.

BCL SLUI Toolkit
Spoken Language User Interface Toolkit (SLUI TK) is in the final stages of development. This project is funded under an Advanced Technology Program (ATP) of NIST for analyzing and understanding SL/NL commands, in conjunction with a standard ASR. The toolkit allows programmers to build an interface that easily integrates speech recognition into new or existing applications. The SLUI attaches to a program, allowing C2 and Logistic Decision Support using spoken NL English, such as:

"What is the current status of Operation Octagon?"
"Send message alpha two to all sector commanders in the battlefield."

The Toolkit handles syntax and semantic processing with minimal direction from the programmer, and removes the need for an in depth understanding of linguistics. Using the SLUI Toolkit, a programmer is able to create a system that incorporates Natural Language Processing (NLP), which includes complex syntactic parsing, and semantic understanding. The SLUI Tool Kit works in the following steps:

1. Toolkit begins to create a SLUI by using NLP to create semantic representations of sample input sentences provided by the programmer.
2. These representations are expanded using synonym sets and other linguistic devices, and stored in a Semantic Frame Table (SFT). The SFT becomes a comprehensive database of all the possible commands a user could request a system to do.
3. The Toolkit then creates methods for attaching the SLUI to the program.
4. When the SLUI enabled program is released, a user may enter a NL sentence. The sentence is translated into a semantic frame, and the SFT is searched for an equivalent frame. If a match is found, the program executes the action linked to this frame.

For more information about BCL SLUI Toolkit, click here

BCL VoiceCommand
Commercial Off The Shelf (COTS) ASRs, such as ViaVoice, are capable of recognizing voice inputs in both C2 and Dictation modes. However, these ASR engines are only capable of understanding voice input if the input matches pre-trained samples. BCL VoiceCommand makes the voice input free from any restrictions by combining the functionalities of the SLUI Toolkit with the functionalities of ASRs using BCL Spoken Language (SL) API Generator.

BCL Papins
Sponsored by DARPA, BCL has developed a portable, efficient, 'principle-and-parameters' based Universal Voice Command Interface System called "Papins" that can operate and control computers and machinery with spoken English. By using an off-the-shelf voice recognition system and a Government and Binding (GB) Theory based natural language parser, BCL can demonstrate the feasibility of developing a system that can be ported from one machine to another by switching the lexicon, GB parameters, and table of machine executable commands.

BCL APA (Adaptive Personal Assistant)
Sponsored by DARPA, BCL developed an Adaptive Personal Assistant (APA) to assist professionals dramatically improve productivity of office automation tasks. The APA is built on an agent-based architecture. A primary agent interfaces with the user, translating natural language requests into the agent languages of KQML and KIF, which are then sent to task-specific agents to be executed. On request completion, each agent shows results in its own display, and a message is sent back to the primary agent for response in a computer-generated voice.

BCL SLGUI (Spoken Language Generic User Interface)
Sponsored by USAF Rome Labs, BCL investigated Spoken Language input for a Generic User Interface (SLGUI) to computer systems. The BCL SLGUI system serves as an infrastructure that allows end-users to use normal day-to-day spoken language to interact with existing off-the-shelf applications with minimal training. In its Phase I work, after investigating how BCL's existing spoken language technology can be integrated with the Air Force's research in generic user interfaces under the GESIA, PESKI, and IeDEA projects, BCL demonstrated a prototype version of the SLGUI. The goal of the demonstration is to explore the feasibility of incorporating spoken language input into existing Air Force applications such as the Interactive Datawall and the Virtual Reality Helmet.

 

1. Extending A Broad-Coverage Parser for a General NLP Toolkit.
   H. Alam, H. Cheng, R. Hartono, A. Kumar, P. Llido, C. Nakatsu, F. Rahman, Y. Tarnikova, T. Tjahjadi, C. Wilcox.
   19th Int. Conf. on Computational Linguistics (COLING 2002), Taipei, Taiwan, pages 22-28.
   http://www.coling2002.sinica.edu.tw/i-Proceedings.html
    
2. Automatic Semantic Grouping in SLUI Toolkit.
   H. Alam, H. Cheng, R. Hartono, A. Kumar, P. Llido, C. Nakatsu, F. Rahman, Y. Tarnikova, T. Tjahjadi, C. Wilcox.
   19th Int. Conf. on Computational Linguistics (COLING 2002), Taipei, Taiwan, pages 15-21.
   http://www.coling2002.sinica.edu.tw/i-Proceedings.html
    
3. Fusion of Two Parsers for a Natural Language Processing Toolkit.
   H. Alam, A. Rahman, T. Tjahjadi, H. Cheng, P. Llido, A. Kumar, R. Hartono, Y. Tarnikova, and C. Wilcox.
   Int. Conf. on Information Fusion, pages 228-234, Maryland, 2002. USA.
   http://www.fusion2002.org/pdf_files/2002_Fusion_Final_Program.pdf
    
4. Development of Spoken Language User Interfaces: A Tool Kit Approach.
   H. Alam, A. Rahman, T. Tjahjadi, H. Cheng, P. Llido, A. Kumar, R. Hartono, Y. Tarnikova, and C. Wilcox.
   Lecture Notes in Computer Science LNCS 2396, T. Caelli, A. Amin, R. Duin, M. Kamel and D. Ridder (Eds.), pages 339-347, 2002.
   http://link.springer.de/link/service/series/0558/tocs/t2396.htm
    
5. A computational linguistic analysis of Bangla using the GB theory
   Z. Khan and R. Berwick
   In Proc. Int. Conf. on Computational Linguistics, Speech and Document Processing, ICCLSDP'98, pp. C1-C8, 1998.
   http://www.isical.ac.in/~cvpr/confer.html

Patent Pending

1. Generating thematic roles in a spoken language user interface toolkit. Provisional Patent, 60/427,394, filed on November 18th, 2002.
2. A comparison of a manual and a statistical approach for the grammar induction for shallow paring. Provisional Patent, 60/427462, filed on November 18th, 2002.
3. BCL VoiceContact. Provisional Patent, 60/417,002, filed on October 7th, 2002.
4. BCL Voice MP3. Provisional Patent, 60/417,029, filed on October 7th, 2002.
5. Voice control for music players. Provisional Patent, 60/399,480, filed on July 29th, 2002.
6. BCL VoiceCommand. Provisional Patent, 60/367,650, Filed on March 25th, 2002.
7. A spoken language user interface toolkit. Provisional Patent 60/354,733, filed on February 4th, 2002.
8. Spoken Language User Interface Toolkit, Provisional Patent, filed on December 18th, 2001.
9. Extending a broad coverage parser (MiniPar) by combining it with a domain-specific Shallow parser for the SLUI Toolkit. Provisional Patent 60/357,734, filed on February 15th, 2002.
10. Rule based semantic tagging with a constrained partial verb shallow parser based on the Levin 1993 classes. Provisional Patent 60/354,758, filed on February 2nd, 2002.
11. Automatic semantic grouping of SLUI toolkit. Provisional Patent, 60/354,734 filed on February 2nd, 2002.
12. Meeting Manager NL processor. Provisional Patent, 60/335,200, filed on October 24, 2001.
13. Adaptive natural language interface. Provisional Patent, 60/404,725, filed on August 19th, 2002.
14. System and Method for Distributed Scheduling. Provisional Patent, filed on November 2nd, 2000.