Search

Single channel speech enhancement is a challenging task in speech community. Recently, various neural networks based methods have been applied to speech enhancement. Among these models, PHASEN and T-GSA achieve state-of-the-art performances on the publicly opened VoiceBank+DEMAND corpus. Both of the models reach the COVL score of 3.62. PHASEN achieves the highest CSIG score of 4.21 while T-GSA gets the highest PESQ score of 3.06. However, both of these two models are very large. The contradiction between the model performance and the model size is hard to reconcile. In this paper, we introduce three kinds of techniques to shrink the PHASEN model and improve the performance. Firstly, seperable polling attention is proposed to replace the frequency transformation blocks in PHASEN. Secondly, global layer normalization followed with PReLU is used to replace batch normalization followed with ReLU. Finally, BLSTM in PHASEN is replaced with Conv2d operation and the phase stream is simplified. With all these modifications, the size of the PHASEN model is shrunk from 33M parameters to 5M parameters, while the performance on VoiceBank+DEMAND is improved to the CSIG score of 4.30, the PESQ score of 3.07 and the COVL score of 3.73.

The era of the automobile has seriously degraded the quality of urban life through costly travel and visible environmental effects. A new urban planning paradigm must be at the heart of our roadmap for the years to come. The one where, within minutes, inhabitants can access their basic living needs by bike or by foot. In this work, we present novel insights of the interplay between the distributions of facilities and population that maximize accessibility over the existing road networks. Results in six cities reveal that travel costs could be reduced in half through redistributing facilities. In the optimal scenario, the average travel distance can be modeled as a functional form of the number of facilities and the population density. As an application of this finding, it is possible to estimate the number of facilities needed for reaching a desired average travel distance given the population distribution in a city.

It becomes urgent to design effective anti-spoofing algorithms for vulnerable automatic speaker verification systems due to the advancement of high-quality playback devices. Current studies mainly treat anti-spoofing as a binary classification problem between bonafide and spoofed utterances, while lack of indistinguishable samples makes it difficult to train a robust spoofing detector. In this paper, we argue that for anti-spoofing, it needs more attention for indistinguishable samples over easily-classified ones in the modeling process, to make correct discrimination a top priority. Therefore, to mitigate the data discrepancy between training and inference, we propose D3M, to leverage a balanced focal loss function as the training objective to dynamically scale the loss based on the traits of the sample itself. Besides, in the experiments, we select three kinds of features that contain both magnitude-based and phase-based information to form complementary and informative features. Experimental results on the ASVspoof2019 dataset demonstrate the superiority of the proposed methods by comparison between our systems and top-performing ones. Systems trained with the balanced focal loss perform significantly better than conventional cross-entropy loss. With complementary features, our fusion system with only three kinds of features outperforms other systems containing five or more complex single models by 22.5% for min-tDCF and 7% for EER, achieving a min-tDCF and an EER of 0.0124 and 0.55% respectively. Furthermore, we present and discuss the evaluation results on real replay data apart from the simulated ASVspoof2019 data, indicating that research for anti-spoofing still has a long way to go. Source code, analysis data, and other details are publicly available at https://github.com/asvspoof/D3M., Comment: The 25th International Conference on Pattern Recognition (ICPR2020)

Nowadays, 23% of the world population lives in multi-million cities. In these metropolises, criminal activity is much higher and violent than in either small cities or rural areas. Thus, understanding what factors influence urban crime in big cities is a pressing need. Mainstream studies analyse crime records through historical panel data or analysis of historical patterns combined with ecological factor and exploratory mapping. More recently, machine learning methods have provided informed crime prediction over time. However, previous studies have focused on a single city at a time, considering only a limited number of factors (such as socio-economical characteristics) and often at large spatial units. Hence, our understanding of the factors influencing crime across cultures and cities is very limited. Here we propose a Bayesian model to explore how crime is related not only to socio-economic factors but also to the built environmental (e.g. land use) and mobility characteristics of neighbourhoods. To that end, we integrate multiple open data sources with mobile phone traces and compare how the different factors correlate with crime in diverse cities, namely Boston, Bogot\'a, Los Angeles and Chicago. We find that the combined use of socio-economic conditions, mobility information and physical characteristics of the neighbourhood effectively explain the emergence of crime, and improve the performance of the traditional approaches. However, we show that the socio-ecological factors of neighbourhoods relate to crime very differently from one city to another. Thus there is clearly no "one fits all" model.

Travel time estimation is a crucial task for not only personal travel scheduling but also city planning. Previous methods focus on modeling toward road segments or sub-paths, then summing up for a final prediction, which have been recently replaced by deep neural models with end-to-end training. Usually, these methods are based on explicit feature representations, including spatio-temporal features, traffic states, etc. Here, we argue that the local traffic condition is closely tied up with the land-use and built environment, i.e., metro stations, arterial roads, intersections, commercial area, residential area, and etc, yet the relation is time-varying and too complicated to model explicitly and efficiently. Thus, this paper proposes an end-to-end multi-task deep neural model, named Deep Image to Time (DeepI2T), to learn the travel time mainly from the built environment images, a.k.a. the morphological layout images, and showoff the new state-of-the-art performance on real-world datasets in two cities. Moreover, our model is designed to tackle both path-aware and path-blind scenarios in the testing phase. This work opens up new opportunities of using the publicly available morphological layout images as considerable information in multiple geography-related smart city applications., Comment: Accepted at IJCAI 2019

Sentiment analysis, mostly based on text, has been rapidly developing in the last decade and has attracted widespread attention in both academia and industry. However, the information in the real world usually comes from multiple modalities, such as audio and text. Therefore, in this paper, based on audio and text, we consider the task of multimodal sentiment analysis and propose a novel fusion strategy including both multi-feature fusion and multi-modality fusion to improve the accuracy of audio-text sentiment analysis. We call it the DFF-ATMF (Deep Feature Fusion - Audio and Text Modality Fusion) model, which consists of two parallel branches, the audio modality based branch and the text modality based branch. Its core mechanisms are the fusion of multiple feature vectors and multiple modality attention. Experiments on the CMU-MOSI dataset and the recently released CMU-MOSEI dataset, both collected from YouTube for sentiment analysis, show the very competitive results of our DFF-ATMF model. Furthermore, by virtue of attention weight distribution heatmaps, we also demonstrate the deep features learned by using DFF-ATMF are complementary to each other and robust. Surprisingly, DFF-ATMF also achieves new state-of-the-art results on the IEMOCAP dataset, indicating that the proposed fusion strategy also has a good generalization ability for multimodal emotion recognition., Comment: Accepted by AAAI2020 Workshop: AffCon2020