Speech recognition systems on the Cell Broadband Engine processor

Abstract

In this paper we describe our design, implementation, and initial results of a prototype connected-phoneme-based speech recognition system on the Cell Broadband Engine™ (Cell/B.E.) processor. Automated speech recognition decodes speech samples into plaintext (other representations are possible) and must process samples at real-time rates. Fortunately, the computational tasks involved in this pipeline are highly data parallel and can receive significant hardware acceleration from vector-streaming architectures such as the Cell/B.E. Architecture. Identifying and exploiting these parallelism opportunities is challenging and critical to improving system performance. From our initial performance timings, we observed that a single Cell/B.E. processor can recognize speech from thousands of simultaneous voice channels in real time-a channel density that is orders of magnitude greater than the capacity of existing software speech recognizers based on CPUs (central processing units). This result emphasizes the potential for Cell/B.E. processor-based speech recognition and will likely lead to the development of production speech systems using Cell/B.E. processor clusters.