Child psychiatrists, pediatricians, and other child clinicians need to have a solid understanding of child language development. There are at least four important reasons that make this necessary. First, slowing, arrest, and deviation of language development are highly associated with, and complicate the course of, child psychopathology. Second, language competence plays a crucial role in emotional and mood regulation, evaluation, and therapy. Third, language deficits are the most frequent underpinning of the learning disorders, ubiquitous in our clinical populations. Fourth, clinicians should not confuse the rich linguistic and dialectal diversity of our clinical populations with abnormalities in child language development. The challenge for the clinician becomes, then, how to get immersed in the captivating field of child language acquisition without getting overwhelmed by its conceptual and empirical complexity. In the past 50 years and since the seminal works of Roger Brown, Jerome Bruner, and Catherine Snow, child language researchers (often known as developmental psycholinguists) have produced a remarkable body of knowledge. Linguists such as Chomsky and philosophers such as Grice have strongly influenced the science of child language. One of the major tenets of Chomskian linguistics (known as generative grammar) is that children’s capacity to acquire language is “hardwired” with “universal grammar”—an innate language acquisition device (LAD), a language “instinct”—at its core. This view is in part supported by the assertion that the linguistic input that children receive is relatively dismal and of poor quality relative to the high quantity and quality of output that they manage to produce after age 2 and that only an advanced, innate capacity to decode and organize linguistic input can enable them to “get from here (prelinguistic infant) to there (linguistic child).” In “Constructing a Language,” Tomasello presents a contrasting theory of how the child acquires language: It is not a universal grammar that allows for language development. Rather, human cognition universals of communicative needs and vocal-auditory processing result in some language universals, such as nouns and verbs as expressions of reference and predication (p. 19). The author proposes that two sets of cognitive skills resulting from biological/phylogenetic adaptations are fundamental to the ontogenetic origins of language. These sets of inherited cognitive skills are intentionreading on the one hand and pattern-finding, on the other. Intention-reading skills encompass the prelinguistic infant’s capacities to share attention to outside events with other persons, establishing joint attentional frames, to understand other people’s communicative intentions, and to imitate the adult’s communicative intentions (an intersubjective form of imitation that requires symbolic understanding and perspective-taking). Pattern-finding skills include the ability of infants as young as 7 months old to analyze concepts and percepts (most relevant here, auditory or speech percepts) and create concrete or abstract categories that contain analogous items. Tomasello, a most prominent developmental scientist with research foci on child language acquisition and on social cognition and social learning in children and primates, succinctly and clearly introduces the major points of his theory and his views on the origins of language in the initial chapters. In subsequent chapters, he delves into the details by covering most language acquisition domains, namely, word (lexical) learning, syntax, and morphology and conversation, narrative, and extended discourse. Although one of the remaining domains (pragmatics) is at the core of his theory and permeates the text throughout, the relative paucity of passages explicitly devoted to discussing acquisition and proBOOK REVIEWS

Constructing a language: A usage-based theory of language acquisition

Single-channel, speaker-independent speech separation methods have recently seen great progress. However, the accuracy, latency, and computational cost of such methods remain insufficient. The majority of the previous methods have formulated the separation problem through the time-frequency representation of the mixed signal, which has several drawbacks, including the decoupling of the phase and magnitude of the signal, the suboptimality of time-frequency representation for speech separation, and the long latency in calculating the spectrograms. To address these shortcomings, we propose a fully-convolutional time-domain audio separation network (Conv-TasNet), a deep learning framework for end-to-end time-domain speech separation. Conv-TasNet uses a linear encoder to generate a representation of the speech waveform optimized for separating individual speakers. Speaker separation is achieved by applying a set of weighting functions (masks) to the encoder output. The modified encoder representations are then inverted back to the waveforms using a linear decoder. The masks are found using a temporal convolutional network (TCN) consisting of stacked 1-D dilated convolutional blocks, which allows the network to model the long-term dependencies of the speech signal while maintaining a small model size. The proposed Conv-TasNet system significantly outperforms previous time-frequency masking methods in separating two- and three-speaker mixtures. Additionally, Conv-TasNet surpasses several ideal time-frequency magnitude masks in two-speaker speech separation as evaluated by both objective distortion measures and subjective quality assessment by human listeners. Finally, Conv-TasNet has a significantly smaller model size and a shorter minimum latency, making it a suitable solution for both offline and real-time speech separation applications.

/pdf/conv-tasnet-surpassing-ideal-time-frequency-magnitude-3fbpvuuk37.pdf

Conv-TasNet: Surpassing Ideal Time-Frequency Magnitude Masking for Speech Separation

Speech separation is the task of separating target speech from background interference. Traditionally, speech separation is studied as a signal processing problem. A more recent approach formulates speech separation as a supervised learning problem, where the discriminative patterns of speech, speakers, and background noise are learned from training data. Over the past decade, many supervised separation algorithms have been put forward. In particular, the recent introduction of deep learning to supervised speech separation has dramatically accelerated progress and boosted separation performance. This paper provides a comprehensive overview of the research on deep learning based supervised speech separation in the last several years. We first introduce the background of speech separation and the formulation of supervised separation. Then, we discuss three main components of supervised separation: learning machines, training targets, and acoustic features. Much of the overview is on separation algorithms where we review monaural methods, including speech enhancement (speech-nonspeech separation), speaker separation (multitalker separation), and speech dereverberation, as well as multimicrophone techniques. The important issue of generalization, unique to supervised learning, is discussed. This overview provides a historical perspective on how advances are made. In addition, we discuss a number of conceptual issues, including what constitutes the target source.

Supervised Speech Separation Based on Deep Learning: An Overview

Recent studies in deep learning-based speech separation have proven the superiority of time-domain approaches to conventional time-frequency-based methods. Unlike the time-frequency domain approaches, the time-domain separation systems often receive input sequences consisting of a huge number of time steps, which introduces challenges for modeling extremely long sequences. Conventional recurrent neural networks (RNNs) are not effective for modeling such long sequences due to optimization difficulties, while one-dimensional convolutional neural networks (1-D CNNs) cannot perform utterance-level sequence modeling when its receptive field is smaller than the sequence length. In this paper, we propose dual-path recurrent neural network (DPRNN), a simple yet effective method for organizing RNN layers in a deep structure to model extremely long sequences. DPRNN splits the long sequential input into smaller chunks and applies intra- and inter-chunk operations iteratively, where the input length can be made proportional to the square root of the original sequence length in each operation. Experiments show that by replacing 1-D CNN with DPRNN and apply sample-level modeling in the time-domain audio separation network (TasNet), a new state-of-the-art performance on WSJ0-2mix is achieved with a 20 times smaller model than the previous best system.

Dual-Path RNN: Efficient Long Sequence Modeling for Time-Domain Single-Channel Speech Separation

Given the recent surge in developments of deep learning, this paper provides a review of the state-of-the-art deep learning techniques for audio signal processing. Speech, music, and environmental sound processing are considered side-by-side, in order to point out similarities and differences between the domains, highlighting general methods, problems, key references, and potential for cross fertilization between areas. The dominant feature representations (in particular, log-mel spectra and raw waveform) and deep learning models are reviewed, including convolutional neural networks, variants of the long short-term memory architecture, as well as more audio-specific neural network models. Subsequently, prominent deep learning application areas are covered, i.e., audio recognition (automatic speech recognition, music information retrieval, environmental sound detection, localization and tracking) and synthesis and transformation (source separation, audio enhancement, generative models for speech, sound, and music synthesis). Finally, key issues and future questions regarding deep learning applied to audio signal processing are identified.

Deep Learning for Audio Signal Processing

Speech enhancement model is used to map a noisy speech to a clean speech. In the training stage, an objective function is often adopted to optimize the model parameters. However, in the existing literature, there is an inconsistency between the model optimization criterion and the evaluation criterion for the enhanced speech. For example, in measuring speech intelligibility, most of the evaluation metric is based on a short-time objective intelligibility (STOI) measure, while the frame based mean square error (MSE) between estimated and clean speech is widely used in optimizing the model. Due to the inconsistency, there is no guarantee that the trained model can provide optimal performance in applications. In this study, we propose an end-to-end utterance-based speech enhancement framework using fully convolutional neural networks (FCN) to reduce the gap between the model optimization and the evaluation criterion. Because of the utterance-based optimization, temporal correlation information of long speech segments, or even at the entire utterance level, can be considered to directly optimize perception-based objective functions. As an example, we implemented the proposed FCN enhancement framework to optimize the STOI measure. Experimental results show that the STOI of a test speech processed by the proposed approach is better than conventional MSE-optimized speech due to the consistency between the training and the evaluation targets. Moreover, by integrating the STOI into model optimization, the intelligibility of human subjects and automatic speech recognition system on the enhanced speech is also substantially improved compared to those generated based on the minimum MSE criterion.

/pdf/end-to-end-waveform-utterance-enhancement-for-direct-11l61fg0uf.pdf

End-to-End Waveform Utterance Enhancement for Direct Evaluation Metrics Optimization by Fully Convolutional Neural Networks

Adversarial loss in a conditional generative adversarial network (GAN) is not designed to directly optimize evaluation metrics of a target task, and thus, may not always guide the generator in a GAN to generate data with improved metric scores. To overcome this issue, we propose a novel MetricGAN approach with an aim to optimize the generator with respect to one or multiple evaluation metrics. Moreover, based on MetricGAN, the metric scores of the generated data can also be arbitrarily specified by users. We tested the proposed MetricGAN on a speech enhancement task, which is particularly suitable to verify the proposed approach because there are multiple metrics measuring different aspects of speech signals. Moreover, these metrics are generally complex and could not be fully optimized by Lp or conventional adversarial losses.

/pdf/metricgan-generative-adversarial-networks-based-black-box-51p6qzbw5o.pdf

MetricGAN: Generative Adversarial Networks based Black-box Metric Scores Optimization for Speech Enhancement.

The electrocardiogram (ECG) is an efficient and noninvasive indicator for arrhythmia detection and prevention. In real-world scenarios, ECG signals are prone to be contaminated with various noises, which may lead to wrong interpretation. Therefore, significant attention has been paid on denoising of ECG for accurate diagnosis and analysis. A denoising autoencoder (DAE) can be applied to reconstruct the clean data from its noisy version. In this paper, a DAE using the fully convolutional network (FCN) is proposed for ECG signal denoising. Meanwhile, the proposed FCN-based DAE can perform compression with regard to the DAE architecture. The proposed approach is applied to ECG signals from the MIT-BIH Arrhythmia database and the added noise signals are obtained from the MIT-BIH Noise Stress Test database. The denoising performance is evaluated using the root-mean-square error (RMSE), percentage-root-mean-square difference (PRD), and improvement in signal-to-noise ratio (SNR
 imp
). The results of the experiments conducted on noisy ECG signals of different levels of input SNR show that the FCN acquires better performance as compared to the deep fully connected neural network- and convolutional neural network-based denoising models. Moreover, the proposed FCN-based DAE reduces the size of the input ECG signals, where the compressed data is 32 times smaller than the original. The results of the study demonstrate the superiority of FCN in denoising, with lower RMSE and PRD, as well as higher SNR
 imp
. According to the results, we believe that the proposed FCN-based DAE has a good application prospect in clinical practice.

/pdf/noise-reduction-in-ecg-signals-using-fully-convolutional-1114x5snpu.pdf

Noise Reduction in ECG Signals Using Fully Convolutional Denoising Autoencoders

This paper proposes a signal-to-noise-ratio (SNR) aware convolutional neural network (CNN) model for speech enhancement (SE). Because the CNN model can deal with local temporal-spectral structures of speech signals, it can effectively disentangle the speech and noise signals given the noisy speech signals. In order to enhance the generalization capability and accuracy, we propose two SNR-aware algorithms for CNN modeling. The first algorithm employs a multi-task learning (MTL) framework, in which restoring clean speech and estimating SNR level are formulated as the main and the secondary tasks, respectively, given the noisy speech input. The second algorithm is an SNR adaptive denoising, in which the SNR level is explicitly predicted in the first step, and then an SNR-dependent CNN model is selected for denoising. Experiments were carried out to test the two SNR-aware algorithms for CNN modeling. Results demonstrate that CNN with the two proposed SNR-aware algorithms outperform the deep neural network counterpart in terms of standardized objective evaluations when using the same number of layers and nodes. Moreover, the SNR-aware algorithms can improve the denoising performance with unseen SNR levels, suggesting their promising generalization capability for real-world applications.

SNR-Aware Convolutional Neural Network Modeling for Speech Enhancement.

This paper aims to address two issues existing in the current speech enhancement methods: 1) the difficulty of phase estimations; 2) a single objective function cannot consider multiple metrics simultaneously. To solve the first problem, we propose a novel convolutional neural network (CNN) model for complex spectrogram enhancement, namely estimating clean real and imaginary (RI) spectrograms from noisy ones. The reconstructed RI spectrograms are directly used to synthesize enhanced speech waveforms. In addition, since log-power spectrogram (LPS) can be represented as a function of RI spectrograms, its reconstruction is also considered as another target. Thus a unified objective function, which combines these two targets (reconstruction of RI spectrograms and LPS), is equivalent to simultaneously optimizing two commonly used objective metrics: segmental signal-to-noise ratio (SSNR) and log-spectral distortion (LSD). Therefore, the learning process is called multi-metrics learning (MML). Experimental results confirm the effectiveness of the proposed CNN with RI spectrograms and MML in terms of improved standardized evaluation metrics on a speech enhancement task.

Szu-Wei Fu

Papers

End-to-End Waveform Utterance Enhancement for Direct Evaluation Metrics Optimization by Fully Convolutional Neural Networks

MetricGAN: Generative Adversarial Networks based Black-box Metric Scores Optimization for Speech Enhancement.

Noise Reduction in ECG Signals Using Fully Convolutional Denoising Autoencoders

SNR-Aware Convolutional Neural Network Modeling for Speech Enhancement.

Complex spectrogram enhancement by convolutional neural network with multi-metrics learning