This paper describes an attempt to extract distinctive phonetic features (DPFs) that represent articulatory gestures in linguistic theory by using a multi-layer neural network (MLN) and to apply the DPFs to noise-robust speech recognition. In the DPF extraction stage, afler converting a speech signal to acoustic features composed of local features (LFs), an MLN with 33 output units corresponding to context-dependent DPFs of I1 DPFs, 11 preceding context DPFs, and I I following context DPFs maps the LFs to DPFs. The proposed DPF parameters without MFCC were firstly evaluated in comparison with a standard parameter set of MFCC and dynamic features on a word recognition task using clean speech and the result showed the same performance as that of the standard set. Noise robustness of these parameters was then tested with four types of additive noise and the proposed DPF parameters outperformed the standard set except one additive noise type.
keywords: causal induction, symmetry bias, mutual exclusivity bias, n-armed bandit problem, trade-off between exploration and exploitation
SummaryThrough numbers of studies on the formation of equivalence relations and causal induction, it is known that human beings tend to consider conditional statements "if p then q" as biconditional statements "if and only if p then q": we call the tendency to perceive "if p then q" as "if q then p" the "symmetry bias". On the other hand, many studies on children's word learning have pointed out that children tend to expect each object has only one label. This is so-called the "mutual exclusivity bias". This bias implies that children infer "if not p then not q" from "if p then q".These biases logically mislead human beings. What is the merit of these illogical induction? In this paper we address this question. First, we clarify the relationship between causal induction and the symmetry bias or and the mutual exclusivity bias. Secondly, we propose a new model of causal induction. Thirdly, we construct an agent which makes illogical decision based on causality, and assess the agent's performance for the task called "n-armed bandit problem" in the field of reinforcement learning. In this problem, it is known that there is a "trade-off between exploration and exploitation". According to our simulation, the agent can resolve the trade-off and achieve quite better performance than an agent without these biases.
SUMMARYThis paper describes a distinctive phonetic feature (DPF) extraction method for use in a phoneme recognition system; our method has a low computation cost. This method comprises three stages. The first stage uses two multilayer neural networks (MLNs): MLN LF−DPF , which maps continuous acoustic features, or local features (LFs), onto discrete DPF features, and MLN Dyn , which constrains the DPF context at the phoneme boundaries. The second stage incorporates inhibition/enhancement (In/En) functionalities to discriminate whether the DPF dynamic patterns of trajectories are convex or concave, where convex patterns are enhanced and concave patterns are inhibited. The third stage decorrelates the DPF vectors using the Gram-Schmidt orthogonalization procedure before feeding them into a hidden Markov model (HMM)-based classifier. In an experiment on Japanese Newspaper Article Sentences (JNAS) utterances, the proposed feature extractor, which incorporates two MLNs and an In/En network, was found to provide a higher phoneme correct rate with fewer mixture components in the HMMs. key words: distinctive phonetic feature, hidden Markov model, multilayer neural network, inhibition/enhancement network, local features
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.