An analytical approach to photonic reservoir computing – a network of SOA's – for noisy speech recognition

This paper seeks to investigate an approach of photonic reservoir computing for optical speech recognition on an examination isolated digit recognition task. An analytical approach in photonic reservoir computing is further drawn on to decrease time consumption, compared to numerical methods; which is very important in processing large signals such as speech recognition. It is also observed that adjusting reservoir parameters along with a good nonlinear mapping of the input signal into the reservoir, analytical approach, would boost recognition accuracy performance. Perfect recognition accuracy (i.e. 100%) can be achieved for noiseless speech signals. For noisy signals with 0–10 db of signal to noise ratios, however, the accuracy ranges observed varied between 92% and 98%. In fact, photonic reservoir application demonstrated 9–18% improvement compared to classical reservoir networks with hyperbolic tangent nodes.