- Award Details
Alternative Transducers for Handheld Automatic Speech Recognition in Military Environments
Department of Defense
Defense Advanced Research Projects Agency
Agency Tracking Number:
Solicitation Topic Code:
Small Business Information
LI CREATIVE TECHNOLOGIES
225 Runnymede Parkway, New Providence, NJ, 07974
Socially and Economically Disadvantaged:
AbstractPhraselator is a handheld speech recognizer/machine translator developed by previous DARPA programs and it is currently in use in Iraq and Afghanistan. As in all existing automatic speech recognition systems, the recognition performances of the Phraselator can be degraded seriously by background military noises and extraneous talking. The objective of this project is to solve the problems and make the Phraselator more useful under military operations. During the Phase I of this project, LcT have successfully completed the feasibility research by introducing a 4-sensor microphone array and an advanced noise reduction/speech enhancement algorithms. Our experiments have showed that we can improve the signal-to-noise (SNR) for about 15 dB. Our solution shows the premise to outperform currently available techniques, and will significant improve the Phraselaor speech recognition accuracy under background noises. During Phase II, we will further improve the algorithms to support two-way machine translation, in which one speaker is close to the Phaselator and another one is in a few meters away. We will also improve the microphone array to suppress any extraneous talking. All the algorithms plus the microphone array will be implemented as hardware in an audio module and integrated into the Phraselator. We will utilize the advanced semiconductor technology, such as mixed-signal chips, for the purpose of developing reliable, light weigh, and low cost hardware/product for immediate applications. Voxtec has agreed to be a subcontractor to take care of the interface design and system integration for a quick deliver to the military use. During Phase III, the technology will be extended to commercial products as the core technique for hi-clean microphones, hearing aids, satellite radio, teleconference equipment, high-quality codec for wireless and landline communications, etc.
* information listed above is at the time of submission.