Dimitra Emmanouilidou

Microsoft Research (United Kingdom)

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Trends

Author Order

Document Type

Co-Authors

Ivan Tashev

Microsoft (United States)

Hannes Gamper

Microsoft (United States)

Mounya Elhilali

Johns Hopkins University

Franco Chiarugi

Foundation for Research and Technology Hellas

Andrew D. Wilson

Microsoft (United States)

Benjamin Elizalde

Microsoft (United States)

Sebastian Braun

Microsoft (United States)

James E. West

Johns Hopkins University

Eric D. McCollum

Johns Hopkins University

Ovidio Salvetti

Istituto di Scienza e Tecnologie dell'Informazione "Alessandro Faedo"

Cooperative Institutions

Johns Hopkins University

Microsoft (United States)

Microsoft Research (United Kingdom)

University of the Witwatersrand

Johns Hopkins Medicine

Centers for Disease Control and Prevention

University of Crete

University of Washington

Center for Global Health

Foundation for Research and Technology Hellas

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Field

Design and Comparative Performance of a Robust Lung Auscultation System for Noisy Clinical Settings

IEEE Journal of Biomedical and Health Informatics (2021)

Ian McLane Dimitra Emmanouilidou J. E. West Mounya Elhilali

Chest auscultation is a widely used clinical tool for respiratory disease detection. The stethoscope has undergone a number of transformative enhancements since its invention, including the introduction of electronic systems in the last two decades. Nevertheless, stethoscopes remain riddled with a number of issues that limit their signal quality and diagnostic capability, rendering both traditional and electronic stethoscopes unusable in noisy or non-traditional environments (e.g., emergency rooms, rural clinics, ambulatory vehicles). This work outlines the design and validation of an advanced electronic stethoscope that dramatically reduces external noise contamination through hardware redesign and real-time, dynamic signal processing. The proposed system takes advantage of an acoustic sensor array, an external facing microphone, and on-board processing to perform adaptive noise suppression. The proposed system is objectively compared to six commercially-available acoustic and electronic devices in varying levels of simulated noisy clinical settings and quantified using two metrics that reflect perceptual audibility and statistical similarity, normalized covariance measure (NCM) and magnitude squared coherence (MSC). The analyses highlight the major limitations of current stethoscopes and the significant improvements the proposed system makes in challenging settings by minimizing both distortion of lung sounds and contamination by ambient noise.

Stethoscope

Auscultation

Robustness

10.1109/jbhi.2021.3056916

Cite

Citations (37)

Supervised Deep Hashing for Efficient Audio Event Retrieval

ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (2020)

Arindam Jati Dimitra Emmanouilidou

Efficient retrieval of audio events can facilitate real-time implementation of numerous query and search-based systems. This work investigates the potency of different hashing techniques for efficient audio event retrieval. Multiple state-of-the-art weak audio embeddings are employed for this purpose. The performance of four classical unsupervised hashing algorithms is explored as part of off-the-shelf analysis. Then, we propose a partially supervised deep hashing framework that transforms the weak embeddings into a low-dimensional space while optimizing for efficient hash codes. The model uses only a fraction of the available labels and is shown here to significantly improve the retrieval accuracy on two widely employed audio event datasets. The extensive analysis and comparison between supervised and unsupervised hashing methods presented here, give insights on the quantizability of audio embeddings. This work provides a first look in efficient audio event retrieval systems and hopes to set baselines for future research.

10.1109/icassp40776.2020.9053766

Cite

Citations (7)

ECG and echocardiography processing for decision support in heart failure

Computers in cardiology (2008)

Franco Chiarugi Sara Colantonio Dimitra Emmanouilidou Davide Moroni Francesco Perticone

This paper presents an effective way to achieve a high level integration of signal and image processing methods in the general management of heart failure, by means of a clinical decision support system (CDSS). In particular, significant and suitably designed image and signal processing algorithms are introduced to objectively and reliably evaluate important features that can facilitate decisional problems in collaboration with the CDSS. Architectural details of the components of the CDSS needed for the seamless integration of image and signal analysis work-flows are finally discussed.

SIGNAL (programming language)

10.1109/cic.2008.4749125

Cite

Citations (2)

Decision support in heart failure through processing of electro- and echocardiograms

Artificial Intelligence in Medicine (2010)

Franco Chiarugi Sara Colantonio Dimitra Emmanouilidou Massimo Martinelli Davide Moroni

10.1016/j.artmed.2010.05.001

Cite

Citations (15)

A Closed-loop Adaptive Brain-computer Interface Framework: Improving the Classifier with the Use of Error-related Potentials

Kuan-Jung Chiang Dimitra Emmanouilidou Hannes Gamper David Johnston Mihai Jalobeanu

Brain-computer interfaces (BCIs) using Electroencephalography (EEG) have drawn attention to providing alternative control pathways for users with motor disabilities or even the general public in real-world environments due to their robustness, relatively low cost, and high portability. However, EEG still suffers from large variability between subjects or between sessions of an individual subject. To obtain optimal performance, a BCI usually requires a user to go through a calibration process to fine-tune the model. This calibration process is usually long and could hinder the practicality of a BCI. In this study, we propose a closed-loop framework that monitors the user EEG responses to the action of a BCI. If an Error-related Potential (ErrP) is detected in the response, it is indicated that the BCI is making a wrong prediction. By using the information from this ErrP detector, we can include online testing trials into the training pool and further fine-tune the model over the time the BCI is used. Results suggest that the proposed framework can reach better results with a few additional trials when compared to the model pre-trained from some existing data. Also, the performance of the proposed model can gradually converge to a fully calibrated model, which suggests that the conventional calibration process could be replaced by online training.

Software portability

Robustness

10.1109/ner49283.2021.9441133

Cite

Citations (9)

Training Audio Captioning Models without Audio

ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (2024)

Soham Deshmukh Benjamin Elizalde Dimitra Emmanouilidou Bhiksha Raj Rita Singh

Automated Audio Captioning (AAC) is the task of generating natural language descriptions given an audio stream. A typical AAC system requires manually curated training data of audio segments and corresponding text caption annotations. The creation of these audio-caption pairs is costly, resulting in general data scarcity for the task. In this work, we address this major limitation and propose an approach to train AAC systems using only text. Our approach leverages the multimodal space of contrastively trained audio-text models, such as CLAP. During training, a decoder generates captions conditioned on the pretrained CLAP text encoder. During inference, the text encoder is replaced with the pretrained CLAP audio encoder. To bridge the modality gap between text and audio embeddings, we propose the use of noise injection or a learnable adapter, during training. We find that the proposed text-only framework performs competitively with stateof-the-art models trained with paired audio, showing that efficient text-to-audio transfer is possible. Finally, we showcase both stylized audio captioning and caption enrichment while training without audio or human-created text captions.

Closed captioning

Audio mining

10.1109/icassp48485.2024.10448115

Cite

Citations (3)

IMPROVING THE QUALITY, ANALYSIS AND INTERPRETATION OF BODY SOUNDS ACQUIRED IN CHALLENGING CLINICAL SETTINGS

Dimitra Emmanouilidou

Source

Cite

Citations (0)

EEGヘッドホンからの音楽注意の解読:ユーザフレンドリーな聴覚脳コンピュータインタフェイス【JST・京大機械翻訳】

IEEE Conference Proceedings (2021)

W An Winko Barbara Shinn‐Cunningham Gamper Hannes Dimitra Emmanouilidou Johnston David

Source

Cite

Citations (0)

Sentiment Detection from Speech Recognition Output

Engineering Sciences (2020)

Iv. Tashev Dimitra Emmanouilidou

Sentiment detection from text has been one of the first text analysis applications.Recently it made serious progress with using deep learning algorithms.Practical use of the sentiment detection includes applications for monitoring the quality of customer service.In this article we perform a review of established and novel features for text analysis, combine them with the latest deep learning algorithms, and evaluate the proposed models.The issues we address are robustness to the low speech recognition rate, the variable length of the text queries, and the case of imbalanced data sets.We use a large labelled dataset from real support calls, and propose new optimality criterion, which is a combination of weighted and unweighted accuracy.

10.7546/engsci.lvii.20.02.01

Cite

Citations (0)

Multi-View Learning for Speech Emotion Recognition with Categorical Emotion, Categorical Sentiment, and Dimensional Scores

ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (2023)

Daniel Tompkins Dimitra Emmanouilidou Soham Deshmukh Benjamin Elizalde

Psychological research has postulated that emotions and sentiment are correlated to dimensional scores of valence, arousal, and dominance. However, the literature of Speech Emotion Recognition focuses on independently predicting the three of them for a given speech audio. In this paper, we evaluate and quantify the predictive power of the dimensional scores towards categorical emotions and sentiment for two publicly available speech emotion datasets. We utilize the three emotional views in a joined multi-view training framework. The views comprise the dimensional scores, emotions categories, and sentiment categories. We present a comparison for each emotional view or combination of, utilizing two general-purpose models for speech-related applications: CNN14 and Wav2Vec2. To our knowledge this is the first time such a joint framework is explored. We found that a joined multi-view training framework can produce results as strong or stronger than models trained independently for each view.

Categorical variable

Sentiment Analysis

Predictive power

Emotion detection

Dominance (genetics)

10.1109/icassp49357.2023.10096700

Cite

Citations (3)