Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

The CHiME challenge series aims to advance far field speech recognition technology by promoting research at the interface of signal processing and automatic speech recognition. This paper presents the design and outcomes of the 3rd CHiME Challenge, which targets the performance of automatic speech recognition in a real-world, commercially-motivated scenario: a person talking to a tablet device that has been fitted with a six-channel microphone array. The paper describes the data collection, the task definition and the baseline systems for data simulation, enhancement and recognition. The paper then presents an overview of the 26 systems that were submitted to the challenge focusing on the strategies that proved to be most successful relative to the MVDR array processing and DNN acoustic modeling reference system. Challenge findings related to the role of simulated data in system training and evaluation are discussed.

/pdf/the-third-chime-speech-separation-and-recognition-challenge-4108e6hzs1.pdf

The third ‘CHiME’ speech separation and recognition challenge: Dataset, task and baselines

We introduce TUT Acoustic Scenes 2016 database for environmental sound research, consisting of binaural recordings from 15 different acoustic environments. A subset of this database, called TUT Sound Events 2016, contains annotations for individual sound events, specifically created for sound event detection. TUT Sound Events 2016 consists of residential area and home environments, and is manually annotated to mark onset, offset and label of sound events. In this paper we present the recording and annotation procedure, the database content, a recommended cross-validation setup and performance of supervised acoustic scene classification system and event detection baseline system using mel frequency cepstral coefficients and Gaussian mixture models. The database is publicly released to provide support for algorithm development and common ground for comparison of different techniques.

TUT database for acoustic scene classification and sound event detection

This paper presents and discusses various metrics proposed for evaluation of polyphonic sound event detection systems used in realistic situations where there are typically multiple sound sources active simultaneously The system output in this case contains overlapping events, marked as multiple sounds detected as being active at the same time The polyphonic system output requires a suitable procedure for evaluation against a reference Metrics from neighboring fields such as speech recognition and speaker diarization can be used, but they need to be partially redefined to deal with the overlapping events We present a review of the most common metrics in the field and the way they are adapted and interpreted in the polyphonic case We discuss segment-based and event-based definitions of each metric and explain the consequences of instance-based and class-based averaging using a case study In parallel, we provide a toolbox containing implementations of presented metrics

Metrics for Polyphonic Sound Event Detection

Sound events often occur in unstructured environments where they exhibit wide variations in their frequency content and temporal structure. Convolutional neural networks CNNs are able to extract higher level features that are invariant to local spectral and temporal variations. Recurrent neural networks RNNs are powerful in learning the longer term temporal context in the audio signals. CNNs and RNNs as classifiers have recently shown improved performances over established methods in various sound recognition tasks. We combine these two approaches in a convolutional recurrent neural network CRNN and apply it on a polyphonic sound event detection task. We compare the performance of the proposed CRNN method with CNN, RNN, and other established methods, and observe a considerable improvement for four different datasets consisting of everyday sound events.

Convolutional Recurrent Neural Networks for Polyphonic Sound Event Detection

This paper presents a system aiming at joint dereverberation and noise reduction by applying a combination of a beamformer with a single-channel spectral enhancement scheme. First, a minimum variance distortionless response beamformer with an online estimated noise coherence matrix is used to suppress noise and reverberation. The output of this beamformer is then processed by a single-channel spectral enhancement scheme, based on statistical room acoustics, minimum statistics, and temporal cepstrum smoothing, to suppress residual noise and reverberation. The evaluation is conducted using the REVERB challenge corpus, designed to evaluate speech enhancement algorithms in the presence of both reverberation and noise. The proposed system is evaluated using instrumental speech quality measures, the performance of an automatic speech recognition system, and a subjective evaluation of the speech quality based on a MUSHRA test. The performance achieved by beamforming, single-channel spectral enhancement, and their combination are compared, and experimental results show that the proposed system is effective in suppressing both reverberation and noise while improving the speech quality. The achieved improvements are particularly significant in conditions with high reverberation times.

/pdf/combination-of-mvdr-beamforming-and-single-channel-spectral-ux0wrt9hz7.pdf

Combination of MVDR beamforming and single-channel spectral processing for enhancing noisy and reverberant speech

Increase in the number of older people due to demographic changes poses great challenges to the social healthcare systems both in the Western and as well as in the Eastern countries. Support for older people by formal care givers leads to enormous temporal and personal efforts. Therefore, one of the most important goals is to increase the efficiency and effectiveness of today’s care. This can be achieved by the use of assistive technologies. These technologies are able to increase the safety of patients or to reduce the time needed for tasks that do not relate to direct interaction between the care giver and the patient. Motivated by this goal, this contribution focuses on applications of acoustic technologies to support users and care givers in ambient assisted living (AAL) scenarios. Acoustic sensors are small, unobtrusive and can be added to already existing care or living environments easily. The information gathered by the acoustic sensors can be analyzed to calculate the position of the user by localization and the context by detection and classification of acoustic events in the captured acoustic signal. By doing this, possibly dangerous situations like falls, screams or an increased amount of coughs can be detected and appropriate actions can be initialized by an intelligent autonomous system for the acoustic monitoring of older persons. The proposed system is able to reduce the false alarm rate compared to other existing and commercially available approaches that basically rely only on the acoustic level. This is due to the fact that it explicitly distinguishes between the various acoustic events and provides information on the type of emergency that has taken place. Furthermore, the position of the acoustic event can be determined as contextual information by the system that uses only the acoustic signal. By this, the position of the user is known even if she or he does not wear a localization device such as a radio-frequency identification (RFID) tag. Category: Smart and intelligent computing

/pdf/acoustic-monitoring-and-localization-for-social-care-5f4x82d89l.pdf

Acoustic Monitoring and Localization for Social Care

A system to estimate the occupancy level of rooms and buildings solely based on acoustic features will be presented in this paper. It is the goal to use this occupancy estimate to increase the energy-efficiency of modern buildings. An overview of the underlying hardware and software concept as well as a brief description of the acoustic occupancy level estimation is given. The demonstrator shows a working prototype which was developed as part of the EU FP7 project - Sounds for Energy Control of Buildings (S4ECoB) [1].

Networked embedded acoustic processing system for smart building applications

In recent years artificial neural networks are successfully applied especially in the context of automatic speech recognition. As information processing systems, neural networks are trained by, e.g., backpropagation or restricted Boltzmann machines to classify patterns at the input of the system. The current work presents the implementation of a deep neural network (DNN) architecture for acoustic source localization.

/pdf/on-sound-source-localization-of-speech-signals-using-deep-32w8jwy3tv.pdf

On sound source localization of speech signals using deep neural networks

The paper describes an automatic speech recognition (ASR) system for the 3rd CHiME challenge that addresses noisy acoustic scenes within public environments. The proposed system includes a multi-channel speech enhancement front-end including a microphone channel failure detection method that is based on cross-comparing the modulation spectra of speech to detect erroneous microphone recordings. The main focus of the submission is the investigation of the amplitude modulation filter bank (AMFB) as a method to extract long-term acoustic cues prior to a Gaussian mixture model (GMM) or deep neural network (DNN) based ASR classifier. It is shown that AMFB features outperform the commonly used frame splicing technique of filter bank features even on a performance optimized ASR challenge system. I.e., temporal analysis of speech by hand-crafted and auditory motivated AMFBs is shown to be more robust compared to a data-driven method based on extracting temporal dynamics with a DNN. Our final ASR system, which additionally includes adaptation of acoustic features to speaker characteristics, achieves an absolute word error rate reduction of approx. 21.53 % relative to the best CHiME-3 baseline system on the "real" test condition.

Stephan Gerlach

Papers

Combination of MVDR beamforming and single-channel spectral processing for enhancing noisy and reverberant speech

Acoustic Monitoring and Localization for Social Care

Networked embedded acoustic processing system for smart building applications

On sound source localization of speech signals using deep neural networks

A CHiME-3 challenge system: Long-term acoustic features for noise robust automatic speech recognition