{"id":3949,"date":"2025-07-04T12:49:34","date_gmt":"2025-07-04T10:49:34","guid":{"rendered":"https:\/\/aholab.ehu.eus\/aholab\/?p=3949"},"modified":"2025-09-19T12:56:22","modified_gmt":"2025-09-19T10:56:22","slug":"mariana-flores-personalized-text-to-speech-voice-generation-for-mexican-spanish-using-neural-tts-models","status":"publish","type":"post","link":"https:\/\/aholab.ehu.eus\/aholab\/mariana-flores-personalized-text-to-speech-voice-generation-for-mexican-spanish-using-neural-tts-models\/","title":{"rendered":"Mariana Flores: Personalized Text-to-Speech Voice Generation for Mexican Spanish Using Neural TTS Models"},"content":{"rendered":"\n<p><strong>Student:<\/strong> Mariana FLores<br><strong>Advisors:<\/strong> Dr. Eva Navas Cord\u00f3n, PhD, Dr. Inma Hernaez Rioja, PhD<br><strong>Thesis defense date: <\/strong>04\/07\/2025<\/p>\n\n\n\n<p>This thesis focuses on developing a personalized Mexican Spanish voice for the AhoMyTTS platform. Leveraging the VITS architecture, multiple training strategies and voice configurations were explored, including the creation and curation of Mexican Spanish corpora and the adaptation of phonetic resources. The project evaluated whether existing modules for Peninsular Spanish could be extended or adapted to Mexican Spanish, and compared the effectiveness of models trained on different language varieties. Results demonstrate that models trained exclusively on Mexican Spanish data best capture the phonetic and prosodic features of the dialect, resulting in improved synthesis quality and naturalness. The methods and resources developed provide a practical foundation for extending TTS personalization to other Spanish varieties.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Student: Mariana FLoresAdvisors: Dr. Eva Navas Cord\u00f3n, PhD, Dr. Inma Hernaez Rioja, PhDThesis defense date: 04\/07\/2025 This thesis focuses on developing a personalized Mexican Spanish voice for the AhoMyTTS platform. Leveraging the VITS architecture, multiple training strategies and voice configurations were explored, including the creation and curation of Mexican Spanish corpora and the adaptation of&#8230;<\/p>\n","protected":false},"author":3,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_es_post_content":"<!-- wp:paragraph -->\n<p><strong>Estudiante:<\/strong> Mariana FLores<br><strong>Directoras:<\/strong> Dr. Eva Navas Cord\u00f3n, PhD, Dr. Inma Hernaez Rioja, PhD<br><strong>Fecha de defensa <\/strong>04\/07\/2025<\/p>\n<!-- \/wp:paragraph -->\n\n<!-- wp:paragraph -->\n<p>Esta tesis se centra en el desarrollo de una voz personalizada en espa\u00f1ol mexicano para la plataforma AhoMyTTS. Aprovechando la arquitectura VITS, se exploraron diversas estrategias de entrenamiento y configuraciones de voz, incluyendo la creaci\u00f3n y curaci\u00f3n de corpus en espa\u00f1ol mexicano y la adaptaci\u00f3n de recursos fon\u00e9ticos.<br>El proyecto evalu\u00f3 si los m\u00f3dulos existentes para el espa\u00f1ol peninsular pod\u00edan extenderse o adaptarse al espa\u00f1ol mexicano, y compar\u00f3 la eficacia de modelos entrenados con distintas variedades del idioma. Los resultados demuestran que los modelos entrenados exclusivamente con datos en espa\u00f1ol mexicano capturan mejor las caracter\u00edsticas fon\u00e9ticas y pros\u00f3dicas del dialecto, lo que se traduce en una mayor calidad y naturalidad en la s\u00edntesis.<br>Los m\u00e9todos y recursos desarrollados ofrecen una base pr\u00e1ctica para extender la personalizaci\u00f3n de sistemas TTS a otras variedades del espa\u00f1ol.<\/p>\n<!-- \/wp:paragraph -->","_es_post_name":"","_es_post_excerpt":"","_es_post_title":"Mariana Flores: Personalized Text-to-Speech Voice Generation for Mexican Spanish Using Neural TTS Models","_eu_post_content":"<!-- wp:paragraph -->\n<p><strong>Ikaslea<\/strong> Mariana FLores<br><strong>Zuzendariak:<\/strong> Dr. Eva Navas Cord\u00f3n, PhD, Dr. Inma Hernaez Rioja, PhD<br><strong>Defentsa-data: <\/strong>04\/07\/2025<\/p>\n<!-- \/wp:paragraph -->\n\n<!-- wp:paragraph -->\n<p>Tesi honek AhoMyTTS plataformarako Mexikoko espainierazko ahots pertsonalizatu baten garapena du ardatz. VITS arkitekturaz baliatuz, ahots-entrenamendu eta -konfigurazioko hainbat estrategia aztertu ziren, Mexikoko gaztelaniazko corpusak sortzea eta ontzea eta baliabide fonetikoak egokitzea barne.<br>Proiektuak penintsulako espainierarako zeuden moduluak Mexikoko espainierara zabaldu edo egokitu zitezkeen ebaluatu zuen, eta hizkuntzaren aldaera ezberdinekin entrenatutako ereduen eraginkortasuna alderatu zuen. Emaitzek erakusten dutenez, Mexikoko espainierazko datuekin soilik entrenatutako ereduek hobeto harrapatzen dituzte euskalkiaren ezaugarri fonetiko eta prosodikoak, eta horrek kalitate eta naturaltasun handiagoa dakar sintesian.<br>Garatutako metodo eta baliabideek oinarri praktikoa eskaintzen dute TTS sistemen pertsonalizazioa espainieraren beste barietate batzuetara hedatzeko.<\/p>\n<!-- \/wp:paragraph -->","_eu_post_name":"","_eu_post_excerpt":"","_eu_post_title":"Mariana Flores: Personalized Text-to-Speech Voice Generation for Mexican Spanish Using Neural TTS Models","_en_post_content":"<!-- wp:paragraph -->\n<p><strong>Student:<\/strong> Mariana FLores<br><strong>Advisors:<\/strong> Dr. Eva Navas Cord\u00f3n, PhD, Dr. Inma Hernaez Rioja, PhD<br><strong>Thesis defense date: <\/strong>04\/07\/2025<\/p>\n<!-- \/wp:paragraph -->\n\n<!-- wp:paragraph -->\n<p>This thesis focuses on developing a personalized Mexican Spanish voice for the AhoMyTTS platform. Leveraging the VITS architecture, multiple training strategies and voice configurations were explored, including the creation and curation of Mexican Spanish corpora and the adaptation of phonetic resources. The project evaluated whether existing modules for Peninsular Spanish could be extended or adapted to Mexican Spanish, and compared the effectiveness of models trained on different language varieties. Results demonstrate that models trained exclusively on Mexican Spanish data best capture the phonetic and prosodic features of the dialect, resulting in improved synthesis quality and naturalness. The methods and resources developed provide a practical foundation for extending TTS personalization to other Spanish varieties.<\/p>\n<!-- \/wp:paragraph -->","_en_post_name":"mariana-flores-personalized-text-to-speech-voice-generation-for-mexican-spanish-using-neural-tts-models","_en_post_excerpt":"","_en_post_title":"Mariana Flores: Personalized Text-to-Speech Voice Generation for Mexican Spanish Using Neural TTS Models","edit_language":"en","footnotes":""},"categories":[62],"tags":[],"class_list":["post-3949","post","type-post","status-publish","format-standard","hentry","category-master-thesis-finished"],"_links":{"self":[{"href":"https:\/\/aholab.ehu.eus\/aholab\/wp-json\/wp\/v2\/posts\/3949","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/aholab.ehu.eus\/aholab\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/aholab.ehu.eus\/aholab\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/aholab.ehu.eus\/aholab\/wp-json\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/aholab.ehu.eus\/aholab\/wp-json\/wp\/v2\/comments?post=3949"}],"version-history":[{"count":1,"href":"https:\/\/aholab.ehu.eus\/aholab\/wp-json\/wp\/v2\/posts\/3949\/revisions"}],"predecessor-version":[{"id":3950,"href":"https:\/\/aholab.ehu.eus\/aholab\/wp-json\/wp\/v2\/posts\/3949\/revisions\/3950"}],"wp:attachment":[{"href":"https:\/\/aholab.ehu.eus\/aholab\/wp-json\/wp\/v2\/media?parent=3949"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/aholab.ehu.eus\/aholab\/wp-json\/wp\/v2\/categories?post=3949"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/aholab.ehu.eus\/aholab\/wp-json\/wp\/v2\/tags?post=3949"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}