arXiv subjects

Shinji Watanabe

Publications and source records attributed to Shinji Watanabe.

At least 505 records · Page 28Linked to original sources

arXiv2018

Effects of Crystalline Electronic Field and Onsite Interorbital Interaction in Yb-based Quasicrystal and Approximant Crystal

To get an insight into a new type of quantum critical phenomena recently discovered in the quasicrystal Yb$_{15}$Al$_{34}$Au$_{51}$ and approximant crystal (AC) Yb$_{14}$Al$_{35}$Au$_{51}$ under pressure, we discuss the property of the crystalline electronic field (CEF) at Yb in the AC and show that uneven CEF levels at each Yb site can appear because of the Al/Au mixed sites. Then we construct the minimal model for the electronic state on the AC by introducing the onsite Coulomb repulsion between the 4f and 5d orbitals at Yb. Numerical calculation for the ground state shows that the lattice constant dependence of the Yb valence well explains the recent measurement done by systematic substitution of elements of Al and Au in the quasicrystal and AC, where the quasicrystal Yb$_{15}$Al$_{34}$Au$_{51}$ is just located at the point from where the Yb-valence starts to change drastically. Our calculation convincingly demonstrates that this is indeed the evidence that this material is just located at the quantum critical point of the Yb-valence transition.

cond-mat.str-el↗

arXiv2018

ESPnet: End-to-End Speech Processing Toolkit

This paper introduces a new open source platform for end-to-end speech processing named ESPnet. ESPnet mainly focuses on end-to-end automatic speech recognition (ASR), and adopts widely-used dynamic neural network toolkits, Chainer and PyTorch, as a main deep learning engine. ESPnet also follows the Kaldi ASR toolkit style for data processing, feature extraction/format, and recipes to provide a complete setup for speech recognition and other speech processing experiments. This paper explains a major architecture of this software platform, several important functionalities, which differentiate ESPnet from other open source ASR toolkits, and experimental results with major ASR benchmarks.

cs.CL↗

arXiv2018

The fifth 'CHiME' Speech Separation and Recognition Challenge: Dataset, task and baselines

The CHiME challenge series aims to advance robust automatic speech recognition (ASR) technology by promoting research at the interface of speech and language processing, signal processing , and machine learning. This paper introduces the 5th CHiME Challenge, which considers the task of distant multi-microphone conversational ASR in real home environments. Speech material was elicited using a dinner party scenario with efforts taken to capture data that is representative of natural conversational speech and recorded by 6 Kinect microphone arrays and 4 binaural microphone pairs. The challenge features a single-array track and a multiple-array track and, for each track, distinct rankings will be produced for systems focusing on robustness with respect to distant-microphone capture vs. systems attempting to address all aspects of the task including conversational language modeling. We discuss the rationale for the challenge and provide a detailed description of the data collection procedure, the task, and the baseline systems for array synchronization, speech enhancement, and conventional and end-to-end ASR.

cs.SD↗

arXiv2018

Student-Teacher Learning for BLSTM Mask-based Speech Enhancement

Spectral mask estimation using bidirectional long short-term memory (BLSTM) neural networks has been widely used in various speech enhancement applications, and it has achieved great success when it is applied to multichannel enhancement techniques with a mask-based beamformer. However, when these masks are used for single channel speech enhancement they severely distort the speech signal and make them unsuitable for speech recognition. This paper proposes a student-teacher learning paradigm for single channel speech enhancement. The beamformed signal from multichannel enhancement is given as input to the teacher network to obtain soft masks. An additional cross-entropy loss term with the soft mask target is combined with the original loss, so that the student network with single-channel input is trained to mimic the soft mask obtained with multichannel input through beamforming. Experiments with the CHiME-4 challenge single channel track data shows improvement in ASR performance.

eess.AS↗

arXiv2018

Building state-of-the-art distant speech recognition using the CHiME-4 challenge with a setup of speech enhancement baseline

This paper describes a new baseline system for automatic speech recognition (ASR) in the CHiME-4 challenge to promote the development of noisy ASR in speech processing communities by providing 1) state-of-the-art system with a simplified single system comparable to the complicated top systems in the challenge, 2) publicly available and reproducible recipe through the main repository in the Kaldi speech recognition toolkit. The proposed system adopts generalized eigenvalue beamforming with bidirectional long short-term memory (LSTM) mask estimation. We also propose to use a time delay neural network (TDNN) based on the lattice-free version of the maximum mutual information (LF-MMI) trained with augmented all six microphones plus the enhanced data after beamforming. Finally, we use a LSTM language model for lattice and n-best re-scoring. The final system achieved 2.74\% WER for the real test set in the 6-channel track, which corresponds to the 2nd place in the challenge. In addition, the proposed baseline recipe includes four different speech enhancement measures, short-time objective intelligibility measure (STOI), extended STOI (eSTOI), perceptual evaluation of speech quality (PESQ) and speech distortion ratio (SDR) for the simulation test set. Thus, the recipe also provides an experimental platform for speech enhancement studies with these performance measures.

cs.SD↗

arXiv2017

Deep Long Short-Term Memory Adaptive Beamforming Networks For Multichannel Robust Speech Recognition

Far-field speech recognition in noisy and reverberant conditions remains a challenging problem despite recent deep learning breakthroughs. This problem is commonly addressed by acquiring a speech signal from multiple microphones and performing beamforming over them. In this paper, we propose to use a recurrent neural network with long short-term memory (LSTM) architecture to adaptively estimate real-time beamforming filter coefficients to cope with non-stationary environmental noise and dynamic nature of source and microphones positions which results in a set of timevarying room impulse responses. The LSTM adaptive beamformer is jointly trained with a deep LSTM acoustic model to predict senone labels. Further, we use hidden units in the deep LSTM acoustic model to assist in predicting the beamforming filter coefficients. The proposed system achieves 7.97% absolute gain over baseline systems with no beamforming on CHiME-3 real evaluation set.

eess.AS↗

arXiv2017

Advances in Joint CTC-Attention based End-to-End Speech Recognition with a Deep CNN Encoder and RNN-LM

We present a state-of-the-art end-to-end Automatic Speech Recognition (ASR) model. We learn to listen and write characters with a joint Connectionist Temporal Classification (CTC) and attention-based encoder-decoder network. The encoder is a deep Convolutional Neural Network (CNN) based on the VGG network. The CTC network sits on top of the encoder and is jointly trained with the attention-based decoder. During the beam search process, we combine the CTC predictions, the attention-based decoder predictions and a separately trained LSTM language model. We achieve a 5-10\% error reduction compared to prior systems on spontaneous Japanese and Chinese speech, and our end-to-end model beats out traditional hybrid ASR systems.

cs.CL↗

arXiv2017

Multichannel End-to-end Speech Recognition

The field of speech recognition is in the midst of a paradigm shift: end-to-end neural networks are challenging the dominance of hidden Markov models as a core technology. Using an attention mechanism in a recurrent encoder-decoder architecture solves the dynamic time alignment problem, allowing joint end-to-end training of the acoustic and language modeling components. In this paper we extend the end-to-end framework to encompass microphone array signal processing for noise suppression and speech enhancement within the acoustic encoding network. This allows the beamforming components to be optimized jointly within the recognition architecture to improve the end-to-end speech recognition objective. Experiments on the noisy speech benchmarks (CHiME-4 and AMI) show that our multichannel end-to-end system outperformed the attention-based baseline with input from a conventional adaptive beamformer.

cs.SD↗

arXiv2017

New Quantum Criticality Revealed under Pressure

Unconventional quantum critical phenomena observed in Yb-based periodic crystals such as YbRh$_2$Si$_2$ and $β$-YbAlB$_4$ have been one of the central issues in strongly correlated electron systems. The common criticality has been discovered in quasicrystal Yb$_{15}$Au$_{51}$Al$_{34}$, which surprisingly persists under pressure at least up to $P=1.5$ GPa. The $T/H$ scaling where the magnetic susceptibility can be expressed as a single scaling function of the ratio of temperature $T$ and magnetic field $H$ has been discovered in the quasicrystal, which is essentially the same as that observed in $β$-YbAlB$_4$. Recently, the $T/H$ scaling as well as the common criticality has been also observed even in the approximant crystal Yb$_{14}$Au$_{51}$Al$_{35}$ under pressure. The theory of critical Yb-valence fluctuation gives a natural explanation for these striking phenomena in a unified way.

cond-mat.str-el↗

arXiv2017

Joint CTC-Attention based End-to-End Speech Recognition using Multi-task Learning

Recently, there has been an increasing interest in end-to-end speech recognition that directly transcribes speech to text without any predefined alignments. One approach is the attention-based encoder-decoder framework that learns a mapping between variable-length input and output sequences in one step using a purely data-driven method. The attention model has often been shown to improve the performance over another end-to-end approach, the Connectionist Temporal Classification (CTC), mainly because it explicitly uses the history of the target character without any conditional independence assumptions. However, we observed that the performance of the attention has shown poor results in noisy condition and is hard to learn in the initial training stage with long input sequences. This is because the attention model is too flexible to predict proper alignments in such cases due to the lack of left-to-right constraints as used in CTC. This paper presents a novel method for end-to-end speech recognition to improve robustness and achieve fast convergence by using a joint CTC-attention model within the multi-task learning framework, thereby mitigating the alignment issue. An experiment on the WSJ and CHiME-4 tasks demonstrates its advantages over both the CTC and attention-based encoder-decoder baselines, showing 5.4-14.6% relative improvements in Character Error Rate (CER).

cs.CL↗

arXiv2016

Single-Channel Multi-Speaker Separation using Deep Clustering

Deep clustering is a recently introduced deep learning architecture that uses discriminatively trained embeddings as the basis for clustering. It was recently applied to spectrogram segmentation, resulting in impressive results on speaker-independent multi-speaker separation. In this paper we extend the baseline system with an end-to-end signal approximation objective that greatly improves performance on a challenging speech separation. We first significantly improve upon the baseline system performance by incorporating better regularization, larger temporal context, and a deeper architecture, culminating in an overall improvement in signal to distortion ratio (SDR) of 10.3 dB compared to the baseline of 6.0 dB for two-speaker separation, as well as a 7.1 dB SDR improvement for three-speaker separation. We then extend the model to incorporate an enhancement layer to refine the signal estimates, and perform end-to-end training through both the clustering and enhancement stages to maximize signal fidelity. We evaluate the results using automatic speech recognition. The new signal approximation objective, combined with end-to-end training, produces unprecedented performance, reducing the word error rate (WER) from 89.1% down to 30.8%. This represents a major advancement towards solving the cocktail party problem.

cs.LG↗

arXiv2016

Origin of Quantum Criticality in Yb-Al-Au Approximant Crystal and Quasicrystal

To get insight into the mechanism of emergence of unconventional quantum criticality observed in quasicrystal Yb$_{15}$Al$_{34}$Au$_{51}$, the approximant crystal Yb$_{14}$Al$_{35}$Au$_{51}$ is analyzed theoretically. By constructing a minimal model for the approximant crystal, the heavy quasiparticle band is shown to emerge near the Fermi level because of strong correlation of 4f electrons at Yb. We find that charge-transfer mode between 4f electron at Yb on the 3rd shell and 3p electron at Al on the 4th shell in Tsai-type cluster is considerably enhanced with almost flat momentum dependence. The mode-coupling theory shows that magnetic as well as valence susceptibility exhibits $χ\sim T^{-0.5}$ for zero-field limit and is expressed as a single scaling function of the ratio of temperature to magnetic field $T/B$ over four decades even in the approximant crystal when some condition is satisfied by varying parameters, e.g., by applying pressure. The key origin is clarified to be due to the strong locality of the critical Yb-valence fluctuation and small Brillouin zone reflecting the large unit cell, giving rise to the extremely-small characteristic energy scale. This also gives a natural explanation for the quantum criticality in the quasicrystal corresponding to the infinite limit of the unit-cell size.

cond-mat.str-el↗

arXiv2016

Basic Properties of Conductivity and Normal Hall Effect in the Periodic Anderson Model

Exact formulas of diagonal conductivity $σ_{xx}$ and Hall conductivity $σ_{xy}$ are derived from the Kubo formula in hybridized two-orbital systems with arbitrary band dispersions. On the basis of the theoretical framework for the Fermi liquid based on these formulas, the ground-state properties of the periodic Anderson model with electron correlation and weak impurity scattering are studied on the square lattice. It is shown that imbalance of the mass-renormalization factors in $σ_{xx}$ and $σ_{xy}$ causes remarkable increase in the valence-fluctuation regime as the f level increases while the cancellation of the renormalization factors causes slight increase in $σ_{xx}$ and $σ_{xy}$ in the Kondo regime. The Hall coefficient $R_{\rm H}$ shows almost constant behavior in both the regimes. Near half filling, $R_{\rm H}$ is expressed by the total hole density as $R_{\rm H}=1/(\bar{n}_{\rm hole}e)$ while $R_{\rm H}$ approaches zero near quarter filling, which reflects the curvature of the Fermi surface. These results hold as far as the damping rate for f electrons is less than about $10~\%$ of the renormalized hybridization gap. From these results we discuss pressure dependence of residual resistivity and normal Hall effect in Ce- and Yb-based heavy electron systems.

cond-mat.str-el↗

arXiv2015

Deep clustering: Discriminative embeddings for segmentation and separation

We address the problem of acoustic source separation in a deep learning framework we call "deep clustering." Rather than directly estimating signals or masking functions, we train a deep network to produce spectrogram embeddings that are discriminative for partition labels given in training data. Previous deep network approaches provide great advantages in terms of learning power and speed, but previously it has been unclear how to use them to separate signals in a class-independent way. In contrast, spectral clustering approaches are flexible with respect to the classes and number of items to be segmented, but it has been unclear how to leverage the learning power and speed of deep networks. To obtain the best of both worlds, we use an objective function that to train embeddings that yield a low-rank approximation to an ideal pairwise affinity matrix, in a class-independent way. This avoids the high cost of spectral factorization and instead produces compact clusters that are amenable to simple clustering methods. The segmentations are therefore implicitly encoded in the embeddings, and can be "decoded" by clustering. Preliminary experiments show that the proposed method can separate speech: when trained on spectrogram features containing mixtures of two speakers, and tested on mixtures of a held-out set of speakers, it can infer masking functions that improve signal quality by around 6dB. We show that the model can generalize to three-speaker mixtures despite training only on two-speaker mixtures. The framework can be used without class labels, and therefore has the potential to be trained on a diverse set of sound types, and to generalize to novel sources. We hope that future work will lead to segmentation of arbitrary sounds, with extensions to microphone array methods as well as image segmentation and other domains.

cs.NE↗

arXiv2014

Wide Quantum Critical Region of Valence Fluctuations: Origin of Robust Quantum Criticality in Quasicrystal Yb15Al34Au51 under Pressure

The mechanism of the emergence of robust quantum criticality in the heavy-electron quasicrystal Yb15Al34Au51 is analyzed theoretically. By constructing a minimal model for the quasicrystal and its crystalline approximant, which contain concentric shell structures with Yb and Al-Au clusters, we show that a set of quantum critical points of the first-order valence transition of Yb appears as spots in the ground-state phase diagram. Their critical regions overlap each other, giving rise to a wide quantum critical region. This well explains the robust criticality observed in Yb15Al34Au51 under pressure, and predicts the emergence of the common criticality in the crystalline approximant under pressure. The wider critical region in the quasicrystal than that in the crystalline approximant in the T-P phase diagram and the field-induced valence-crossover "region" in the T-H phase diagram are predicted to appear.

cond-mat.str-el↗

arXiv2014

T/B Scaling in beta-YbAlB4

The scaling behavior over four decades of the ratio of temperature T to magnetic field B observed in the magnetization in beta-YbAlB4 is theoretically examined. By developing a theoretical framework that exhibits the quantum critical phenomena of Yb-valence fluctuations under a magnetic field, we show that the T/B-scaling behavior can appear near the quantum critical point of the valence transition. The emergence of the T/B scaling indicates the presence of the small characteristic energy scale of critical Yb-valence fluctuations. It is argued that the quantum valence criticality offers a unified explanation for the unconventional quantum criticality as well as the T/B scaling in beta-YbAlB4.

cond-mat.str-el↗

arXiv2014

Unconventional Quantum Criticality due to Critical Valence Transition

Quantum criticality due to the valence transition in some Yb-based heavy fermion metals has gradually turned out to play a crucial role to understand the non-Fermi liquid properties that cannot be understood from the conventional quantum criticality theory due to magnetic transitions. Namely, critical exponents giving the temperature (T) dependence of the resistivity ρ(T), the Sommerfeld coefficient, C(T)/T, the magnetic susceptibility, χ(T), and the NMR relaxation rates, 1/(T_{1}T), can be understood as the effect of the critical valence fluctuations of f electrons in Yb ion in a unified way. There also exist a series of Ce-based heavy fermion metals that exhibit anomalies in physical quantities, enhancements of the residual resistivity ρ_{0} and the superconducting critical temperature (T_c) around the pressure where the valence of Ce sharply changes. Here we review the present status of these problems both from experimental and theoretical aspects.

cond-mat.str-el↗

arXiv2013

Robustness of Quantum Criticality of Valence Fluctuations

The mechanism of emergence of robust quantum criticality in Yb- and Ce-based heavy electron systems under pressure is analyzed theoretically. By constructing a minimal model for quasicrystal Yb15Al34Au51 and its approximant, we show that quantum critical points of the first-order valence transition of Yb appear in the ground-state phase diagram with their critical regimes being overlapped to be unified, giving rise to a wide quantum critical regime. This well explains the robust unconventional criticality observed in Yb15Al34Au51 under pressure. We also discuss broader applicability of this mechanism to other Yb- and Ce-based systems such as beta-YbAlB4 showing unconventional quantum criticality.

cond-mat.str-el↗