<link>https://kgrosero.github.io/</link></image><item><title>ChiReSSD</h1> <article> <h1>ChiReSSD</h1> <p>Tue, 01 Dec 2026 00:00:00 +0000</p> <p>Speaker-preserving reconstruction of disordered child speech, built on StyleTTS2. Synthesizes a target transcript with canonical pronunciation in the child’s own voice.</p> </article> <article> <h1>Generative Reconstruction of Pediatric Disordered Speech for Automated Clinical Evaluation</h1> <p>Tue, 01 Dec 2026 00:00:00 +0000</p> </article> <article> <h1>MMDyS</h1> <p>Thu, 01 Oct 2026 00:00:00 +0000</p> <p>Multimodal dysarthria severity assessment that aligns speech, text and facial asymmetry. Includes pretrained video, audio-text and fusion checkpoints.</p> </article> <article> <h1>Multimodal Biomarkers of Dysarthria: Severity-Conditioned Contrastive Alignment of Speech, Text, and Facial Asymmetry</h1> <p>Thu, 01 Oct 2026 00:00:00 +0000</p> </article> <article> <h1>Experience</h1> <p>Wed, 30 Sep 2026 00:00:00 +0000</p> </article> <article> <h1>Conditioned Atypical Text-to-Speech for Acoustic and Prosodic Fidelity via Flow Matching</h1> <p>Tue, 01 Sep 2026 00:00:00 +0000</p> </article> <article> <h1>PRiSM: Benchmarking Phone Realization in Speech Models</h1> <p>Wed, 01 Jul 2026 00:00:00 +0000</p> </article> <article> <h1>A Cascaded Unsupervised–Supervised NLP Pipeline for Detecting Accusatory Language in Public Procurement</h1> <p>Mon, 01 Jun 2026 00:00:00 +0000</p> </article> <article> <h1>Enhancing SELD Performance: The Role of Data Augmentation Techniques in Spatial Sound Analysis</h1> <p>Fri, 01 May 2026 00:00:00 +0000</p> </article> <article> <h1>Improving Dysarthria Assessment Through Voice Conversion in Low-Resource Settings</h1> <p>Wed, 15 Oct 2025 00:00:00 +0000</p> </article> <article> <h1>Fine-Tuning Wav2Vec2 for Low-Resource Kichwa Automatic Speech Recognition</h1> <p>Tue, 14 Oct 2025 00:00:00 +0000</p> </article> <article> <h1>DiffusionCleft: Facial Anomaly Synthesis Guided by Text</h1> <p>Mon, 13 Oct 2025 00:00:00 +0000</p> </article> <article> <h1>Anonymizing Faces, Preserving Anomalies</h1> <p>Wed, 01 Oct 2025 00:00:00 +0000</p> <p class="text-xl font-semibold">Deep-Learning Based Privacy Protection for Data of Patients with Cleft Lip</p> <p><strong>Karen Rosero</strong>, Lucas M. Harrison, Alex A. Kane, Carlos Busso, Rami R. Hallac<br> <em>15th International Congress on Cleft Lip/Palate and Related Craniofacial Anomalies (CLEFT 2025), Kyoto, Japan</em><br> In collaboration with UT Southwestern Medical Center and Children’s Health.</p> <h2 id="why-anonymize-clinical-photos">Why anonymize clinical photos?</h2> <p>Facial photographs support treatment planning for patients with cleft lip: lip marking, assessment of surgical outcomes, facial symmetry analysis, and automatic detection of lip abnormalities. Using facial images in medical contexts also raises critical ethical and legal concerns about patients’ privacy, which limits how these images can be shared for research and education.</p> <p><figure class="kr-figure"></figure></p> <h2 id="traditional-anonymization-falls-short">Traditional anonymization falls short</h2> <p>Black bars over the eyes, pixelation and blurring are the usual ways to de-identify a face. Generative AI can reconstruct a highly similar face from any of them, so they do not reliably protect the patient.</p> <p><figure class="kr-figure"></figure></p> <p><figure class="kr-figure"></figure></p> <h2 id="our-approach-replace-the-face-keep-the-anomaly">Our approach: replace the face, keep the anomaly</h2> <p>We preserve the orofacial anatomy of the patient and use a generative AI system to synthesize the rest of the face. The result is a realistic portrait of a person who does not exist, with the clinically relevant region left exactly as it was.</p> <p><figure class="kr-figure"></figure></p> <ol> <li><strong>Landmarks.</strong> Facial landmarks are detected automatically on the original image, using our (FG 2024).</li> <li><strong>Mask.</strong> A mask isolates the orofacial region of interest. It can be adjusted manually or personalized for other facial regions.</li> <li><strong>Synthesis.</strong> A diffusion model, guided by the orofacial region and a text prompt (for example, <em>“light skin girl, green eyes”</em>), generates the new face by progressive denoising. The region of interest stays unaltered, and the text controls the appearance of the synthetic face.</li> </ol> <p><figure class="kr-figure"></figure></p> <p><figure class="kr-figure"></figure></p> <h2 id="results">Results</h2> <p>The approach generalizes to both pre- and post-operative cleft lip cases and produces realistic, anatomically consistent synthetic faces.</p> <p><figure class="kr-figure"></figure></p> <p>We anonymized <strong>140 images</strong> of patients with cleft lip, evenly distributed between pre- and post-operative cases, and asked two questions.</p> <p><strong>Is the generated face recognized as the same person?</strong> Each original image was compared with 8 synthesized variants using a deep-learning identity verification tool, for a total of <strong>1,120 generated images</strong>. <strong>98.9%</strong> were classified as distinct individuals. The remaining 1.1% were flagged as potential matches, all with low confidence (below 70%).</p> <p><figure class="kr-figure"></figure></p> <p><strong>Is the orofacial region preserved?</strong> A clinical evaluator rated 50 images on a 1–5 scale, where 5 means high anatomical preservation after anonymization. <strong>48 images received a 5</strong> and the other 2 received a 4.</p> <p><figure class="kr-figure"></figure></p> <h2 id="takeaways">Takeaways</h2> <ul> <li>A privacy-preserving method for facial portraits of patients with cleft lip.</li> <li>It retains the clinical features that matter for evaluation.</li> <li>It makes it possible to build and share more diverse clinical datasets.</li> </ul> <p><em>Images on this page are illustrations, images generated by our method, or original images blurred as in the conference presentation.</em></p> <p><strong>Related work:</strong> (ICMI 2025) · (FG 2024)</p> </article> <article> <h1>Anonymizing Faces, Preserving Anomalies: Deep-Learning Based Privacy Protection for Data of Patients with Cleft Lip</h1> <p>Wed, 01 Oct 2025 00:00:00 +0000</p> </article> <article> <h1>Advancing Pediatric ASR: The Role of Voice Generation in Disordered Speech</h1> <p>Sun, 17 Aug 2025 00:00:00 +0000</p> </article> <article> <h1>CS-FLEURS: A Massively Multilingual and Code-Switched Speech Dataset</h1> <p>Sun, 17 Aug 2025 00:00:00 +0000</p> </article> <article> <h1>Towards Large-Scale Self-Supervised Learning for Spatial Audio: Generating First-Order Ambisonics from Monophonic Recordings</h1> <p>Fri, 01 Aug 2025 00:00:00 +0000</p> </article> <article> <h1>Deep Learning-Based Assessment of Lip Symmetry for Patients with Repaired Cleft Lip</h1> <p>Sat, 01 Feb 2025 00:00:00 +0000</p> </article> <article> <h1>Mixture of Emotion Dependent Experts: Facial Expressions Recognition in Videos Through Stacked Expert Models</h1> <p>Wed, 15 Jan 2025 00:00:00 +0000</p> </article> <article> <h1>w2v-SELD: A Sound Event Localization and Detection Framework for Self-Supervised Spatial Audio Pre-Training</h1> <p>Tue, 03 Dec 2024 00:00:00 +0000</p> </article> <article> <h1>Lip Abnormality Detection for CLP</h1> <p>Mon, 04 Nov 2024 00:00:00 +0000</p> <p>Lip normalization and Siamese abnormality detection for patients with repaired cleft lip and palate, trained without patient data.</p> </article> <article> <h1>Lip Abnormality Detection for Patients with Repaired Cleft Lip and Palate: A Lip Normalization Approach</h1> <p>Mon, 04 Nov 2024 00:00:00 +0000</p> </article> <article> <h1>Enhanced Facial Landmarks Detection for Patients with Repaired Cleft Lip and Palate</h1> <p>Mon, 27 May 2024 00:00:00 +0000</p> </article> <article> <h1>Lips Symmetry Evaluation in Patients with Repaired Cleft Lip and Palate Using Self-Supervised Machine Learning</h1> <p>Mon, 01 Apr 2024 00:00:00 +0000</p> </article> <article> <h1>Learning Feature Representations from Unlabeled Data for Volcano-Seismic Event Classification</h1> <p>Mon, 01 Jan 2024 00:00:00 +0000</p> </article> <article> <h1>A-MAD: An Adaptive-Microseismic Activity Detector Based on Gaussian Mixture Models and Spectral Subtraction</h1> <p>Tue, 12 Dec 2023 00:00:00 +0000</p> </article> <article> <h1>Demystifying Spatial Audio Plugins: Comparative Insights into Functionality and Features</h1> <p>Tue, 10 Oct 2023 00:00:00 +0000</p> </article> <article> <h1>G-SELD</h1> <p>Thu, 08 Jun 2023 00:00:00 +0000</p> <p>Sound event localization and detection in spatial audio with bio-inspired gammatone filters and temporal convolutional networks.</p> </article> <article> <h1>Sound Events Localization and Detection Using Bio-Inspired Gammatone Filters and Temporal Convolutional Neural Networks</h1> <p>Thu, 08 Jun 2023 00:00:00 +0000</p> </article> <article> <h1>Song Emotion Recognition: A Performance Comparison Between Audio Features and Artificial Neural Networks</h1> <p>Sat, 24 Sep 2022 00:00:00 +0000</p> </article> <article> <h1>Song Emotion Recognition: A Study of the State of the Art</h1> <p>Sun, 24 Oct 2021 00:00:00 +0000</p> </article> <article> <h1>Análise e aplicação do modelo auditivo baseado nos coeficientes cepstrais da frequência de Mel</h1> <p>Sun, 01 Nov 2020 00:00:00 +0000</p> </article> </main></body></html>