Qingyang Hong

A genetic classification method for speaker recognition

Gaussian mixture model (GMM) has been widely used for modeling speakers. In speaker identification, one major problem is how to generate a set of GMMs for identification purposes based upon the training data. Due to the hill-climbing characteristic of the maximum likelihood (ML) method, any arbitrary estimate of the initial model parameters will usually lead to a sub-optimal model in practice. To resolve this problem, this paper proposes a hybrid training method based on genetic algorithm (GA). It utilizes the global searching capability of GA and combines the effectiveness of the ML method. Experimental results based on TI46 and TIMIT showed that this hybrid GA could obtain more optimized GMMs and better results than the simple GA and the traditional ML method.

A discriminative training approach for text-independent speaker recognition

Gaussian mixture model (GMM) has been commonly used for text-independent speaker recognition. The estimation of model parameters is generally performed based on the maximum likelihood (ML) criterion. However, this criterion only utilizes the labeled utterances for each speaker model and very likely leads to a local optimization solution. To solve this problem, this paper proposes a discriminative training approach based on the maximum model distance (MMD) criterion. We investigate the characteristics of speaker recognition and further propose a novel selection strategy of competing speakers associated with it. Experimental results based on the KING and TIMIT databases demonstrate that our training approach was quite efficient to improve the performance of speaker identification and verification. When there were three training sentences for each speaker, the verification equal error rate (EER) of 168 speakers in TIMIT could be reduced by 30.4% compared with the conventional method.

Using mel-frequency cepstral coefficients in missing data technique

Filter bank is the most common feature being employed in the research of the marginalisation approaches for robust speech recognition due to its simplicity in detecting the unreliable data in the frequency domain. In this paper, we propose a hybrid approach based on the marginalisation and the soft decision techniques that make use of the Mel-frequency cepstral coefficients (MFCCs) instead of filter bank coefficients. A new technique for estimating the reliability of each cepstral component is also presented. Experimental results show the effectiveness of the proposed approaches.

Fuzzy neural network based dynamic path planning

It is an important issue for mobile robot to find the best route as well as avoid moving into obstacles. In this article, we put forward a solution to the problem by using fuzzy-neural network. Compared with the other path planning approaches, one of the main advantages of the methods based on fuzzy-neural network is that they give stronger robustness to the robot. Different from the similar methods, we introduce a novel fuzzy membership function which is based on collision prediction. This method not only preserves the advantages of the existing ones, but also can give a realistic meaning to the path gotten from this approach. The simulation results prove the feasibility and validity of our method.

Discriminative training for speaker identification based on maximum model distance algorithm

In this paper we apply the maximum model distance (MMD) training to speaker identification and a new selection strategy of competitive speakers is proposed to it. The traditional ML method only utilizes the utterances for each speaker model, which probably leads to a local optimization solution. By maximizing the dissimilarities among those similar speaker models, MMD could add the discriminative capability into the training procedure and then improve the identification performance. Based on the TIMIT corpus, we designed the word and sentence experiments to evaluate this proposed training approach. The results show that the identification performance can be improved greatly when the training data is limited.

Adaptation of hidden Markov models using maximum model distance algorithm

This paper presents a new approach that uses the maximum model distance (MMD) method for the adaptation of hidden Markov models (HMMs). This method has the same framework as it is used for constructing speech recognizers with abundant data, and work effectively with any amount of adaptation data. All parameters of the HMMs with or without the adaptation data could be adapted. If the adaptation data is sufficient, then the adapted models will gradually become a speaker-dependent one. Both the dialect and the speaker adaptation experiments were conducted to investigate the effectiveness of the proposed algorithm. In the speaker adaptation experiments, up to 65.55% phoneme error reduction was achieved, and the MMD could reduce the phoneme error by 16.91% even only one adaptation utterance is available.

A transfer learning method for PLDA-based speaker verification

Currently, the state-of-the-art speaker verification system is based on i-vector and PLDA. However, PLDA requires tens of thousands of development data from many speakers. This makes it difficult to learn the PLDA parameters for a domain with scarce data. In this paper, we propose an effective transfer learning method based on Bayesian joint probability in which Kullback-Leibler (KL) divergence between the source domain and the target domain is added as a regularization factor. This hypothesis could utilize the development data of source domain to help find a better optimal solution of PLDA parameters for the target domain. Experimental results based on the NIST SRE and Switchboard corpus demonstrate that our proposed method could produce the largest gain of performance compared with the traditional PLDA and the other adaptation approach.

Transfer learning for PLDA-based speaker verification

Abstract Currently, the majority of the state-of-the-art speaker verification systems are based on i-vector and PLDA; however, PLDA requires a huge volume of development data from multiple different speakers. This makes it difficult to learn PLDA parameters for a domain with scarce data. In this paper, we study and extend an effective transfer learning method based on Bayesian joint probability, in which the Kullback–Leibler (KL) divergence between the source domain and the target domain is added as a regularization factor. This method utilizes the development data from the source domain to help find the optimal PLDA parameters for the target domain. Specifically, speaker verification of short utterances can be viewed as a task in the domain with a limited amount of long utterances. Therefore, transfer learning for PLDA can also be adopted to learn discriminative information from other domains with a great deal of long utterances. Experimental results based on the NIST SRE and Switchboard corpus demonstrate that the proposed method offers a significant performance gain when compared with the traditional PLDA.

A Robust Speaker-Adaptive and Text-Prompted Speaker Verification System

Currently, the recording playback attack has become a major security risk for speaker verification. The text-independent or text-dependent system is being troubled by it. In this paper, we propose an effective text-prompted system to overcome this problem, in which speaker verification and speech recognition are combined together. We further adopt speaker-adaptive hidden Markov model (HMM) so as to improve the verification performance. After HMM-based speaker adaptation, this system needs not to be retrained at each verification step. Experimental results demonstrated that the proposed method had quite good performance with the equal error rate (EER) lower than 2% and was also robust for different cases.

A model for ranking sentence pairs in parallel corpora

In this paper, the problem of ranking sentence pairs in parallel corpora was addressed for the first time. To solve this problem, a novel model was proposed. In this model, both syntax features and semantics features of sentence pairs are considered. Since most todays statistical machine translation models depend on word alignment, features related to word alignment information are also included. Two experiments were carried out and the results showed that the model had promising performance.