Expanded Speech Recognition to Include Foreign Accents

In this proposal, Intelligent Automation, Incorporated (IAI) and its subcontractors, Prof. Richard Stern of Carnegie Mellon University (CMU) and Dr. Rita Singh of Haikya Corp., propose a novel integrated system to improve speech recognition performance for people with foreign accents. It is emphasized that this team has rich experience in understanding the speech characteristics in non-native speakers. The temporal and intra-phoneme variations introduce mismatches between the baseline speech recognition model and a particular non-native speaker model. Hence some adaptations are needed to update the speaker models for non-native speakers. The proposed system consists of three parts: 1) a speech enhancement module to eliminate background noise; 2) a speaker adaptation algorithm that uses a small training data set to update the speakers’ models in the speech recognition part; 3) a speech recognition system to recognition speech. Our Phase 1 results have clearly demonstrated the feasibility of the individual algorithms. The goal of this Phase 2 research is to develop and demonstrate a real-time prototype speech recognition system for non-native speakers. The system can be portable to any platform such as PowerPC. The system also has continuous and unsupervised learning capability.