Japanese Newspaper Article Sentences Research Articles

Improved Speech Separation Performance from Monaural Mixed Speech Based on Deep Embedding Network

Speech separation refers to the separation of mixed utterances in which multiple people are speaking simultaneously. This topic was created to address the cocktail party problem, and it has also evolved into a front-end procedure for speech-related research recently. There are two types of models for addressing it: classification methods and regression methods. Classification methods avoid considering the permutation problem, and regression methods are more precise. Deep clustering (DC) is a common deep learning classification technique that uses a deep embedding network to embed signals in the underlying manifold and allows signals with similar properties to cluster together in the embedded space. The ideal binary mask (IBM) supervises the model. Even if it is entirely isolated via supervised binary masking, this setting is susceptible to mistake. To remedy this flaw, this paper proposes a network with two modules based on deep embedding. More precise masks, such as the ideal ratio mask (IRM), can now be used instead of IBM. Furthermore, the cascade structure may preserve DC's high performance. On average, the suggested model beats the original DC model by 1.55dB in SNR, 4.45dB in SDR, 4.41dB in SDRi, 0.16 in STOI, and 0.3 in PESQ on a mixture of Japanese Newspaper Article Sentences (JNAS). Finally, a comparison is made with ChimeraNet which has a similar structure, the results prove the proposed model's superiority over previous DC-related studies.

Speaker Identification and Verification by Combining MFCC and Phase Information

In conventional speaker recognition methods based on Mel-frequency cepstral coefficients (MFCCs), phase information has hitherto been ignored. In this paper, we propose a phase information extraction method that normalizes the change variation in the phase according to the frame position of the input speech and combines the phase information with MFCCs in text-independent speaker identification and verification methods. There is a problem with the original phase information extraction method when comparing two phase values. For example, the difference in the two values of π-\mathtildeθ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sub> and \mathtildeθ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> =-π+\mathtildeθ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sub> is 2π-2\mathtildeθ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sub> . If \mathtildeθ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sub> ≈ 0, then the difference ≈ 2π, despite the two phases being very similar to one another. To address this problem, we map the phase into coordinates on a unit circle. Speaker identification and verification experiments are performed using the NTT database which consists of sentences uttered by 35 (22 male and 13 female) Japanese speakers with normal, fast and slow speaking modes during five sessions. Although the phase information-based method performs worse than the MFCC-based method, it augments the MFCC and the combination is useful for speaker recognition. The proposed modified phase information is more robust than the original phase information for all speaking modes. By integrating the modified phase information with the MFCCs, the speaker identification rate was improved to 98.8% from 97.4% (MFCC), and equal error rate for speaker verification was reduced to 0.45% from 0.72% (MFCC), respectively. We also conducted the speaker identification and verification experiments on a large-scale Japanese Newspaper Article Sentences (JNAS) database, a similar trend as NTT database was obtained.

Inhibition/Enhancement Network Based ASR using Multiple DPF Extractors

This paper describes an evaluation ofInhibition/Enhancement (In/En) network for robustautomatic speech recognition (ASR). In distinctive phoneticfeatures (DPFs) based speech recognition using neuralnetwork, In/En network is needed to discriminate whetherthe DPFs dynamic patterns of trajectories are convex orconcave. The network is used to achieve categorical DPFsmovement by enhancing DPFs peak patterns (convexpatterns) and inhibiting DPFs dip patterns (concavepatterns). We have analyzed the effectiveness of In/Enalgorithm by incorporating it into a system which consists ofthree stages: a) Multilayer Neural Networks (MLNs), b)In/En Network and c) Gram-Schmidt (GS)orthogonalization. From the experiments using JapaneseNewspaper Article Sentences (JNAS) database in clean andnoisy acoustic environments, it is observed that the In/Ennetwork plays a significant role on the improvement ofphoneme recognition performance. Moreover, In/Ennetwork reduces required number of mixture componentsin Hidden Markov Models (HMMs).

Speaker Recognition by Combining MFCC and Phase Information in Noisy Conditions

In this paper, we investigate the effectiveness of phase for speaker recognition in noisy conditions and combine the phase information with mel-frequency cepstral coefficients (MFCCs). To date, almost speaker recognition methods are based on MFCCs even in noisy conditions. For MFCCs which dominantly capture vocal tract information, only the magnitude of the Fourier Transform of time-domain speech frames is used and phase information has been ignored. High complement of the phase information and MFCCs is expected because the phase information includes rich voice source information. Furthermore, some researches have reported that phase based feature was robust to noise. In our previous study, a phase information extraction method that normalizes the change variation in the phase depending on the clipping position of the input speech was proposed, and the performance of the combination of the phase information and MFCCs was remarkably better than that of MFCCs. In this paper, we evaluate the robustness of the proposed phase information for speaker identification in noisy conditions. Spectral subtraction, a method skipping frames with low energy/Signal-to-Noise (SN) and noisy speech training models are used to analyze the effect of the phase information and MFCCs in noisy conditions. The NTT database and the JNAS (Japanese Newspaper Article Sentences) database added with stationary/non-stationary noise were used to evaluate our proposed method. MFCCs outperformed the phase information for clean speech. On the other hand, the degradation of the phase information was significantly smaller than that of MFCCs for noisy speech. The individual result of the phase information was even better than that of MFCCs in many cases by clean speech training models. By deleting unreliable frames (frames having low energy/SN), the speaker identification performance was improved significantly. By integrating the phase information with MFCCs, the speaker identification error reduction rate was about 30%-60% compared with the standard MFCC-based method.

Distinctive Phonetic Feature (DPF) Extraction Based on MLNs and Inhibition/Enhancement Network

This 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): MLNLF-DPF, which maps continuous acoustic features, or local features (LFs), onto discrete DPF features, and MLNDyn, 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.

Differences between acoustic characteristics of spontaneous and read speech and their effects on speech recognition performance

Although speech derived from read texts, news broadcasts, and other similar prepared contexts can be recognized with high accuracy, recognition performance drastically decreases for spontaneous speech. This is due to the fact that spontaneous speech and read speech are significantly different acoustically as well as linguistically. This paper statistically and quantitatively analyzes differences in acoustic features between spontaneous and read speech using two large-scale speech corpora, “Corpus of Spontaneous Japanese (CSJ)” and “Japanese Newspaper Article Sentences (JNAS)”. Experimental results show that spontaneous speech can be characterized by reduced spectral space in comparison with that of read speech, and that the more spontaneous, the more the spectral space shrinks. This paper also clarifies that reduction in the spectral space leads to reduction in phoneme recognition accuracy. This result indicates that spectral reduction is one major reason for the decrease of recognition accuracy in spontaneous speech.

Document‐level optimization in speech recognition

Most ASR systems optimize scores at the sentence level because language models are designed to assign a meaningful probability to a sentence, but not to a document. However, in the last half‐decade, NIPS people have specifically examined aspect‐based global language models and developed generative text models such as the latent Dirichlet allocation (LDA), which can assign a meaningful probability to an entire document. We investigated a system to recognize all sentences in a read document, maximizing a global score including the document probability, in addition to assigning acoustic and local language probabilities. Document level optimization using global scores engenders a combinatorial problem in its implementation. Therefore, we used the rescoring framework with N‐best recognition results for each sentence in a read document and the hill‐climbing method to search for approximately optimal recognition results. In this study, we compare mixture of unigrams (MU), Dirichlet mixtures (DM), and LDAs as generative text models in experiments using read document speech data in Japanese Newspaper Article Sentences (JNAS) and academic presentation speech data in The Corpus of Spontaneous Japanese (CSJ).

JNAS: Japanese speech corpus for large vocabulary continuous speech recognition research.

In this paper we present the first public Japanese speech corpus for large vocabulary continuous speech recognition (LVCSR) technology, which we have titled JNAS (Japanese Newspaper Article Sentences). We designed it to be comparable to the corpora used in the American and European LVCSR projects. The corpus contains speech recordings (60 h) and their orthographic transcriptions for 306 speakers (153 males and 153 females) reading excerpts from the newspaper's articles and phonetically balanced (PB) sentences. This corpus contains utterances of about 45, 000 sentences as a whole with each speaker reading about 150 sentences. JNAS is being distributed on 16 CD-ROMs.