Cosme Llerena-Aguilar

A Computationally Efficient Sound Environment Classifier for Hearing Aids

A computationally efficient system for sound environment classification in digital hearing aids is presented in this paper. The goal is to automatically classify three different listening environments: “speech,” “music,” and “noise.” The system is designed considering the computational limitations found in such devices. The proposed algorithm is based on a novel set of heuristically designed features inspired in the Mel frequency cepstral coefficients. Experiments carried out with real signals demonstrate that the three listening environments can be robustly classified with the proposed system, obtaining low error rates when using a small part of the total computational resources of the DSP of the device. This study demonstrates that the proposed system can be implemented with the available resources in state-of-the-art digital hearing aids.

Energy-weighted Mean Shift algorithm for speech source separation

Blind Source Separation algorithms have been applied to speech mixtures during many years, taking into account the knowledge and properties of speech signals. A new approach for speech separation based on sparse representations of speech has recently arisen. These methods are commonly known as Time-Frequency Masking methods, being the most famous the DUET algorithm that performs separation of undetermined mixtures from only two microphones. Sparsity property also encourages the idea of applying clustering techniques for source separation. In this work, we introduce an adapted version of the clustering method Mean Shift for the separation of speech sources. Obtained results confirm the validity of the method for speech separation improving the DUET performance and showing better generalization. Furthermore, the use of clustering techniques for separation enables the automatic identification of the number of sources.

Synchronization based on mixture alignment for sound source separation in wireless acoustic sensor networks

Desynchronization degrades the performance of many signal processing algorithms in Wireless Acoustic Sensor Networks. It is mainly caused by the different distances between the source and each node and by the clock phase offset and frequency skew. Classical solutions use clock synchronization protocols and algorithms in the communication layer, but these alternatives do not tackle the lack of synchronization caused by the distances between sources and nodes.In this paper, we present a novel study of the synchronization problem in acoustic sensor networks from a signal processing point of view. First, we propose a theoretical framework that allows us to study the effects of misalignment over any short-time based algorithm, focusing on the requirements of the effective length of the analysis time frame. From this framework, a theoretical synchronization delay is established aimed at reducing the required length of the time frame. Second, two novel alignment methods are developed and are tuned up to reduce the amount of synchronization information required for transmission. The results obtained demonstrate that our proposed methods represent a good solution in terms of performance over the quality of a standard Blind Source Separation algorithm, allowing us to reduce the transmission bandwidth required for synchronization data. HighlightsWe study the effects of misalignment over BSS algorithms in WASN.A theoretical synchronization relaxes the constraint over the time frame length.Two novel alignment methods are inspired in the theoretical synchronization.The proposed methods synchronize the mixtures with a reduced transmission bandwidth.

A new mixing matrix estimation method based on the geometrical analysis of the sound separation problem

Nowadays, some of the most successful sound source separation methods are based on the assumption of sparse sources. A large number of those separation solutions consist of two parts: the mixing matrix estimation and the separation stages. Concerning the first part, many sparsity-based separation methods rely on the use of clustering techniques to identify the samples of the mixtures due to each sound source. With certain types of sources, such as speech, the assumption of sparsity is questionable and so, these stages do not perform correctly.In this paper, we present a new mixing matrix estimation procedure to overcome sparsity-based methods separating speech sources. Our novel proposal establishes a geometric relationship between the mixing parameters using some available information about the microphone array, such as, number and type of microphones or the distance between them. Using this relationship, the complex estimation of the level differences is avoided. Results demonstrate that our proposal outperforms mixing matrix estimation solutions in terms of both speech separation quality and speech intelligibility. HighlightsSound separation problems are geometrically studied in a novel way.A new mixing matrix estimation method is introduced.A theoretical relationship between time and level differences is determined.Only time differences must be estimated, avoiding the calculation of the level ones.Time differences are obtained with a classical algorithm robust to reverberation.

Distributed and collaborative sound environment information extraction in binaural hearing aids

Current research in the field of Wireless Acoustic Sensor Networks (WASN) is gradually introducing the use of sound spatial techniques in the field of binaural hearing aids, in which sound environment information must be extracted in order to tune up the main hearing aid algorithms. In binaural hearing aids, computational capability, memory and data transmission are strictly constrained, which makes the use of distributed and collaborative approaches suitable. This paper proposes solutions for the collaborative and distributed sound environment information extraction through the estimation of the different noise levels, analyzing both the performance and the computational and transmission requirements. Results demonstrate that the proposed distributed solutions highly reduce the transmission rate and the computational cost, while maintaining the accuracy in the estimations.

TWO-SENSOR EEG-BASED STRESS DETECTION SYSTEM

In this paper, we propose a computationally-efficient EEGbased stress detection that uses only two non-invasive electrodes. The system is designed to classify between two situations: high stress level or low stress level. A linear classifier is trained using supervised learning using a subset of features that has been selected among a larger proposed set of features, using a tailored feature selection algorithm. The proposed algorithm has been evaluated with subjects playing skill games, obtaining errors of 19.2% in the train set and 29.2% in the test set.