With the emergence of multi-modal generative models, synthesized videos are becoming increasingly realistic, making the detection of deepfakes extremely challenging. While several video deepfake detection models have shown promising performance, their focus has been primarily on the visual modality. To overcome this limitation, we propose a dualstream framework that fuses visual and auditory information via cross-attention computed between embeddings extracted from pre-trained video and audio encoders. Additionally, we design a weakly-supervised forgery localization head that infers frame-level forgery scores from coarse segment-level labels, minimizing the need for fine-grained annotations and allowing for forgery location characterization. In this paper, we describe our preliminary results showing the proposed model outperforming state-of-the-art detectors on both frame-level localization and sequence-level deepfake detection tasks. Ongoing work focuses on investigating the complementarity between the visual and auditory modalities to improve model robustness and explainability.

Voice-based diagnostic systems offer a scalable solution for remote health assessment. However, recent advances in generative voice models may enable malicious manipulation of voice samples to simulate or conceal disease-related speech characteristics, which poses new risks to diagnostic systems. This paper investigates the vulnerability of diagnostic and detection models to such types of "deepfake" attacks. We show that it is possible to train a generative model to convert between healthy voices and pathological ones, which in turn, can successfully deceive existing diagnostic systems. Here, focus is placed on COVID-19 infection and respiratory abnormalities, but the method can be applied across different pathological conditions affecting vocal attributes. We also benchmark four state-of-the-art synthesized voice detection models on both real and generated pathological speech from three datasets. Our results show that current synthetic voice detectors, typically trained on healthy speech data, perform poorly on generated pathological samples. While fine-tuning with real pathological voices improves detection, a substantial performance gap remains. This work provides initial insights on an emerging threat to remote voice diagnostic systems that needs further work.

Speech is known to carry health-related attributes, which has emerged as a novel venue for remote and long-term health monitoring. However, existing models are usually tailored for a specific type of disease, and have been shown to lack generalizability across datasets. Furthermore, concerns have been raised recently towards the leakage of speaker identity from health embeddings. To mitigate these limitations, we propose WavRx, a speech health diagnostics model that captures the respiration and articulation related dynamics from a universal speech representation. Our in-domain and cross-domain experiments on six pathological speech datasets demonstrate WavRx as a new state-of-the-art health diagnostic model. Furthermore, we show that the amount of speaker identity entailed in the WavRx health embeddings is significantly reduced without extra guidance during training. An in-depth analysis of the model was performed, thus providing physiological interpretation of its improved generalizability and privacy-preserving ability.

Speech COVID-19 detection systems have gained popularity as they represent an easy-to-use and low-cost solution that is well suited for at-home long-term monitoring of patients with persistent symptoms. Recently, however, the limited generalization capability of existing deep neural network based systems to unseen datasets has been raised as a serious concern, as has their limited interpretability. In this study, we aim to develop an interpretable and generalizable speech-based COVID-19 detection system. First, we propose the use of a 3-dimensional modulation frequency tensor (called modulation tensorgram representation, MTR) as input to a convolutional recurrent neural network for COVID-19 detection. The MTR representation is known to capture long-term dynamics of speech correlated with articulation and respiration, hence being a potential candidate for characterizing COVID-19 speech. The customized network explores both the spectral and temporal pattern from MTR to learn the underlying COVID-19 speech pattern. Next, we design a spectro-temporal saliency masking to aggregate regions of the MTR related to COVID-19, thus helping further improve the generalizability and interpretability of the model. Experiments are conducted on three public datasets and results show the proposed solution consistently outperforming two benchmark systems in within-, across-, and unseen-dataset tests. The learned salient regions have been shown correlated with whispered speech and vocal hoarseness, which explains the increased generalizability. Furthermore, our model relies on a small amount of parameters, thus offering a promising solution for on-device remote monitoring of COVID-19 infection.

With advances seen in deep learning, voice-based applications are burgeoning, ranging from personal assistants, affective computing, to remote disease diagnostics. As the voice contains both linguistic and para-linguistic information (e.g., vocal pitch, intonation, speech rate, loudness), there is growing interest in voice anonymization to preserve speaker privacy and identity. Voice privacy challenges have emerged over the last few years and focus has been placed on removing speaker identity while keeping linguistic content intact. For affective computing and disease monitoring applications, however, the para-linguistic content may be more critical. Unfortunately, the effects that anonymization may have on these systems are still largely unknown. In this paper, we fill this gap and focus on one particular health monitoring application: speech-based COVID-19 diagnosis. We test three anonymization methods and their impact on five different state-of-the-art COVID-19 diagnostic systems using three public datasets. We validate the effectiveness of the anonymization methods, compare their computational complexity, and quantify the impact across different testing scenarios for both within- and across-dataset conditions. Additionally, we provided a comprehensive evaluation of the importance of different speech aspects for diagnostics and showed how they are affected by different types of anonymizers. Lastly, we show the benefits of using anonymized external data as a data augmentation tool to help recover some of the COVID-19 diagnostic accuracy loss seen with anonymization.

Cough is an important symptom of numerous respiratory diseases, including COVID-19. While different cough phases (i.e., inhalation, compression, and expulsion) have been shown to be related to different pathological origins, existing cough-based COVID-19 detection systems rely on the entire cough recording, thus such phase-related characteristics are overlooked. In this study, our aim is two-fold. First, we have annotated over 1,250 cough recordings from two publicly-available cough sound databases, thus providing the research community with fine-grained cough phase labels. Next, we extract a number of temporal and acoustic features from each cough phase and test their usefulness and complementarity for COVID-19 detection. Experiments show the importance of cough phase segmentation, not only for improved COVID-19 detection, but also for the development of models that are interpretable and can better generalize across datasets.

Self-supervised speech pre-training has emerged as a useful tool to extract representations from speech that can be used across different tasks. While these models are starting to appear in commercial systems, their robustness to so-called adversarial attacks have yet to be fully characterized. This paper evaluates the vulnerability of three self-supervised speech representations (wav2vec 2.0, HuBERT and WavLM) to three white-box adversarial attacks under different signal-to-noise ratios (SNR). The study uses keyword spotting as a downstream task and shows that the models are very vulnerable to attacks, even at high SNRs. The paper also investigates the transferability of attacks between models and analyses the generated noise patterns in order to develop more effective defence mechanisms. The modulation spectrum shows to be a potential tool for detection of adversarial attacks to speech systems.

The coronavirus disease 2019 (COVID-19) pandemic has drastically impacted life around the globe. As life returns to pre-pandemic routines, COVID-19 testing has become a key component, assuring that travellers and citizens are free from the disease. Conventional tests can be expensive, time-consuming (results can take up to 48h), and require laboratory testing. Rapid antigen testing, in turn, can generate results within 15-30 minutes and can be done at home, but research shows they achieve very poor sensitivity rates. In this paper, we propose an alternative test based on speech signals recorded at home with a portable device. It has been well-documented that the virus affects many of the speech production systems (e.g., lungs, larynx, and articulators). As such, we propose the use of new modulation spectral features and linear prediction analysis to characterize these changes and design a two-stage COVID-19 prediction system by fusing the proposed features. Experiments with three COVID-19 speech datasets (CSS, DiCOVA2, and Cambridge subset) show that the two-stage feature fusion system outperforms the benchmark systems of CSS and Cambridge datasets while maintaining lower complexity compared to DL-based systems. Furthermore, the two-stage system demonstrates higher generalizability to unseen conditions in a cross-dataset testing evaluation scheme. The generalizability and interpretability of our proposed system demonstrate the potential for accessible, low-cost, at-home COVID-19 testing.

Existing speech-based coronavirus disease 2019 (COVID-19) detection systems provide poor interpretability and limited robustness to unseen data conditions. In this paper, we propose a system to overcome these limitations. In particular, we propose to fuse two different feature modalities with patient metadata in order to capture different properties of the disease. The first feature set is based on modulation spectral properties of speech. The second comprises spectral shape/descriptor features recently used for COVID-19 detection. Lastly, we fuse patient metadata in order to improve robustness and interpretability. Experiments are performed on the 2021 INTERSPEECH COVID Speech Sub-Challenge dataset with several different data partitioning paradigms. Results show the importance of the modulation spectral features. Metadata, in turn, did not perform very well when used alone but provided invaluable insights when fused with the other features. Overall, a system relying on the fusion of all three modalities showed to be robust to unseen conditions and to rely on interpretable features. The simplicity of the model suggests that it can be deployed in portable devices, hence providing accessible COVID-19 diagnostics worldwide.