{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,12]],"date-time":"2026-08-12T13:43:25Z","timestamp":1786542205742,"version":"build-2736575974"},"reference-count":15,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,8,12]],"date-time":"2026-08-12T00:00:00Z","timestamp":1786492800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Digit. Health"],"abstract":"<jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>Current mental health diagnostic methods are limited by subjective clinical interpretation. Automatic speech analysis is a promising technology for objective assessment.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Objective<\/jats:title>\n                    <jats:p>To evaluate and compare different speech-based representations (acoustic, phonetic, and time-frequency) and deep learning-based embeddings for discriminating symptoms associated with psychological distress.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>A secondary analysis of the Distress Analysis Interview Corpus (DAIC-WOZ) was conducted using recordings from 125 participants (3,069 responses). Speech representations included phonation, articulation, and prosody features extracted with DisVoice; phonetic features extracted with Phonet; time-frequency representations derived from Mexican hat wavelets; and deep embeddings extracted with the multilingual Wav2Vec 2.0 model XLSR-53. Two classification strategies were addressed at the response and participant levels using a Fully Connected Neural Network (FCNN) and a Support Vector Machine (SVM), respectively.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>\n                      Prosody at the participant level achieved the highest mean performance (F1-score 0.67\n                      <jats:inline-formula>\n                        <mml:math xmlns:mml=\"http:\/\/www.w3.org\/1998\/Math\/MathML\" id=\"IM1\">\n                          <mml:mo>\u00b1<\/mml:mo>\n                        <\/mml:math>\n                      <\/jats:inline-formula>\n                      0.07; accuracy 0.64\n                      <jats:inline-formula>\n                        <mml:math xmlns:mml=\"http:\/\/www.w3.org\/1998\/Math\/MathML\" id=\"IM2\">\n                          <mml:mo>\u00b1<\/mml:mo>\n                        <\/mml:math>\n                      <\/jats:inline-formula>\n                      0.10; AUC 0.65\n                      <jats:inline-formula>\n                        <mml:math xmlns:mml=\"http:\/\/www.w3.org\/1998\/Math\/MathML\" id=\"IM3\">\n                          <mml:mo>\u00b1<\/mml:mo>\n                        <\/mml:math>\n                      <\/jats:inline-formula>\n                      0.12), followed by participant-level phonation (F1-score 0.59\n                      <jats:inline-formula>\n                        <mml:math xmlns:mml=\"http:\/\/www.w3.org\/1998\/Math\/MathML\" id=\"IM4\">\n                          <mml:mo>\u00b1<\/mml:mo>\n                        <\/mml:math>\n                      <\/jats:inline-formula>\n                      0.16; accuracy 0.61\n                      <jats:inline-formula>\n                        <mml:math xmlns:mml=\"http:\/\/www.w3.org\/1998\/Math\/MathML\" id=\"IM5\">\n                          <mml:mo>\u00b1<\/mml:mo>\n                        <\/mml:math>\n                      <\/jats:inline-formula>\n                      0.14; AUC 0.65\n                      <jats:inline-formula>\n                        <mml:math xmlns:mml=\"http:\/\/www.w3.org\/1998\/Math\/MathML\" id=\"IM6\">\n                          <mml:mo>\u00b1<\/mml:mo>\n                        <\/mml:math>\n                      <\/jats:inline-formula>\n                      0.16). Conversely, participant-level aggregation of deep embeddings yielded lower performance (F1-score 0.48\n                      <jats:inline-formula>\n                        <mml:math xmlns:mml=\"http:\/\/www.w3.org\/1998\/Math\/MathML\" id=\"IM7\">\n                          <mml:mo>\u00b1<\/mml:mo>\n                        <\/mml:math>\n                      <\/jats:inline-formula>\n                      0.19; accuracy 0.55\n                      <jats:inline-formula>\n                        <mml:math xmlns:mml=\"http:\/\/www.w3.org\/1998\/Math\/MathML\" id=\"IM8\">\n                          <mml:mo>\u00b1<\/mml:mo>\n                        <\/mml:math>\n                      <\/jats:inline-formula>\n                      0.13; AUC 0.52\n                      <jats:inline-formula>\n                        <mml:math xmlns:mml=\"http:\/\/www.w3.org\/1998\/Math\/MathML\" id=\"IM9\">\n                          <mml:mo>\u00b1<\/mml:mo>\n                        <\/mml:math>\n                      <\/jats:inline-formula>\n                      0.15), failing to surpass traditional features. Response-level performance remained close to chance. Phonet and wavelet representations did not improve performance over prosody or phonation.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>Participant-level analysis provided more robust and consistent discriminative patterns than response-level approaches. Prosody and phonation achieved the best performance across speech representations, while phonetic, time-frequency, and deep speech representations did not outperform the best acoustic baseline. These findings suggest that, within the evaluated experimental setting, the aggregation strategy appears to have a stronger influence on performance than increasing representational complexity.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fdgth.2026.1886893","type":"journal-article","created":{"date-parts":[[2026,8,12]],"date-time":"2026-08-12T13:26:46Z","timestamp":1786541206000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Automatic analysis of speech representations to assess psychological distress"],"prefix":"10.3389","volume":"8","author":[{"given":"Sara","family":"Fern\u00e1ndez-Velasco","sequence":"first","affiliation":[{"name":"Telematics Department, Universidad del Cauca","place":["Popay\u00e1n, Colombia"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jose","family":"Moreno-Mesa","sequence":"additional","affiliation":[{"name":"GITA Lab, Universidad de Antioquia","place":["Medell\u00edn, Colombia"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Daniel","family":"Escobar-Grisales","sequence":"additional","affiliation":[{"name":"GITA Lab, Universidad de Antioquia","place":["Medell\u00edn, Colombia"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Diego M.","family":"Lopez","sequence":"additional","affiliation":[{"name":"Telematics Department, Universidad del Cauca","place":["Popay\u00e1n, Colombia"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Juan Rafael","family":"Orozco-Arroyave","sequence":"additional","affiliation":[{"name":"GITA Lab, Universidad de Antioquia","place":["Medell\u00edn, Colombia"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,8,12]]},"reference":[{"key":"B1","year":""},{"key":"B2","doi-asserted-by":"publisher","first-page":"100350","DOI":"10.1016\/j.health.2024.100350","article-title":"A comprehensive review of predictive analytics models for mental illness using machine learning algorithms","volume":"6","author":"Islam","year":"2024","journal-title":"Healthc Anal"},{"key":"B3","author":"Scherer","year":""},{"key":"B4","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1016\/j.specom.2019.10.003","article-title":"Automatic depression classification based on affective read sentences: opportunities for text-dependent analysis","volume":"115","author":"Stasak","year":"2019","journal-title":"Speech Commun"},{"key":"B5","doi-asserted-by":"publisher","first-page":"10","DOI":"10.1016\/j.specom.2015.03.004","article-title":"A review of depression and suicide risk assessment using speech analysis","volume":"71","author":"Cummins","year":"2015","journal-title":"Speech Commun"},{"key":"B6","doi-asserted-by":"publisher","first-page":"96","DOI":"10.1002\/lio2.354","article-title":"Automated assessment of psychiatric disorders using speech: a systematic review","volume":"5","author":"Low","year":"2020","journal-title":"Laryngoscope Invest Otolaryngol"},{"key":"B7","author":"Khan","year":""},{"key":"B8","author":"Gratch","year":""},{"key":"B9","author":"Williamson","year":""},{"key":"B10","author":"Al Hanai","year":""},{"key":"B11","doi-asserted-by":"publisher","first-page":"56","DOI":"10.1016\/j.specom.2020.07.005","article-title":"Parallel representation learning for the classification of pathological speech: studies on Parkinson\u2019s disease and cleft lip and palate","volume":"122","author":"V\u00e1squez-Correa","year":"2020","journal-title":"Speech Commun"},{"key":"B12","author":"V\u00e1squez-Correa","year":""},{"key":"B13","doi-asserted-by":"publisher","first-page":"237","DOI":"10.3390\/s26010237","article-title":"A method for explainable epileptic seizure detection through wavelet transforms obtained by electroencephalogram-based audio recordings","volume":"26","author":"Tavolato","year":"2026","journal-title":"Sensors"},{"key":"B14","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1186\/s12888-019-2375-1","article-title":"Longitudinal trajectories of comorbid PTSD and depression symptoms among U.S. service members and veterans","volume":"19","author":"Armenta","year":"2019","journal-title":"BMC Psychiatry"},{"key":"B15","doi-asserted-by":"publisher","first-page":"11155","DOI":"10.1038\/s41598-023-35184-7","article-title":"Manifestation of depression in speech overlaps with characteristics used to represent and recognize speaker identity","volume":"13","author":"Dumpala","year":"2023","journal-title":"Sci Rep"}],"container-title":["Frontiers in Digital Health"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fdgth.2026.1886893\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,12]],"date-time":"2026-08-12T13:26:47Z","timestamp":1786541207000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fdgth.2026.1886893\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,12]]},"references-count":15,"alternative-id":["10.3389\/fdgth.2026.1886893"],"URL":"https:\/\/doi.org\/10.3389\/fdgth.2026.1886893","relation":{},"ISSN":["2673-253X"],"issn-type":[{"value":"2673-253X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,12]]},"article-number":"1886893"}}