{"id":3699,"date":"2024-04-16T14:28:11","date_gmt":"2024-04-16T12:28:11","guid":{"rendered":"https:\/\/aholab.ehu.eus\/aholab\/?p=3699"},"modified":"2024-06-05T13:33:10","modified_gmt":"2024-06-05T11:33:10","slug":"ilenia-disponible-el-sistema-de-reconocimiento-de-voz-en-euskera","status":"publish","type":"post","link":"https:\/\/aholab.ehu.eus\/aholab\/ilenia-disponible-el-sistema-de-reconocimiento-de-voz-en-euskera\/","title":{"rendered":"Basque Speech Recognition System available"},"content":{"rendered":"\n

From the Aholab<\/a> Group and the HiTZ<\/a> center we present to the public the new speech recognition system in Basque. This technological advance has the potential to transform the interaction between people and technology, especially in the field of the Basque language.<\/p>\n\n\n\n

The system has been trained with 548 hours of Basque voices from different public sources (Mozilla Common Voice 16.1<\/a>, Basque Parliament<\/a>, OpenSLR<\/a>), which allows it to accurately recognize the words and phrases spoken by users, reaching quality levels of WER less than 5%.<\/p>\n\n\n\n

Two different models have been created based on NVIDIA pre-trained models. One of them<\/a> using a language model with more classic techniques, and the other<\/a> using more emerging technologies such as transducers. The training of the models was carried out on the Hyperion system from the DIPC servers<\/a>.<\/p>\n\n\n\n

The system can potentially be integrated into virtual assistants to perform tasks such as sending messages, searching for information or setting reminders. It could also enable the automation of responses to telephone calls, improving efficiency and customer service. And it will facilitate the transcription of audio recordings in Basque.<\/p>\n\n\n\n

A demo of the speech recognition system in Basque is available at this link<\/a> and the models are available at Gaitu-Data<\/a>. The team invites the community to use it and provide feedback to continue improving the technology. We hope that it will be a valuable tool for the Basque community and contribute to the strengthening of our language.<\/p>\n","protected":false},"excerpt":{"rendered":"

From the Aholab Group and the HiTZ center we present to the public the new speech recognition system in Basque. This technological advance has the potential to transform the interaction between people and technology, especially in the field of the Basque language. The system has been trained with 548 hours of Basque voices from different…<\/p>\n","protected":false},"author":3,"featured_media":3703,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_es_post_content":"\n

Desde el Grupo Aholab<\/a> y el centro HiTZ<\/a> presentamos al p\u00fablico el nuevo sistema de reconocimiento de voz en euskera<\/strong>. Este avance tecnol\u00f3gico tiene el potencial de transformar la interacci\u00f3n entre las personas y la tecnolog\u00eda, especialmente en el \u00e1mbito del idioma vasco.<\/p>\n\n\n\n

El sistema ha sido entrenado con 548 horas de voz en euskera procedentes de distintas fuentes p\u00fablicas (Mozilla Common Voice 16.1<\/a>, Basque Parliament<\/a>,  OpenSLR<\/a>). lo que le permite reconocer con precisi\u00f3n las palabras y frases habladas por los usuarios, alcanzando niveles de calidad de WER menores al 5%. <\/p>\n\n\n\n

Se han creado dos modelos diferentes basados en modelos preentrenados de NVIDIA. Uno de ellos<\/a> se ha creado utilizando un modelo de lenguaje con t\u00e9cnicas m\u00e1s cl\u00e1sicas, y el otro<\/a> utilizando tecnolog\u00edas m\u00e1s emergentes como los transducers. El entrenamiento de los modelos se ha llevado a cabo en el sistema Hyperion de los servidores del DIPC<\/a>.<\/p>\n\n\n\n

Potencialmente, el sistema puede integrarse en asistentes virtuales para realizar tareas como enviar mensajes, buscar informaci\u00f3n o establecer recordatorios. Tambi\u00e9n podr\u00eda posibilitar la automatizaci\u00f3n de respuestas en llamadas telef\u00f3nicas, mejorando la eficiencia y la atenci\u00f3n al cliente. Y, sin duda, facilitar\u00e1 la transcripci\u00f3n de grabaciones de audio en euskera.<\/p>\n\n\n\n

En este enlace<\/a> est\u00e1 disponible una demo del sistema y los modelos pueden descargarse de Gaitu-Data<\/a>. El equipo invita a la comunidad a utilizarlo y proporcionar comentarios para seguir mejorando la tecnolog\u00eda. Esperamos que sea una herramienta valiosa para la comunidad vasca y contribuya al fortalecimiento de nuestro idioma.<\/p>\n","_es_post_name":"","_es_post_excerpt":"","_es_post_title":"Disponible el Sistema de Reconocimiento de Habla en Euskera","_eu_post_content":"\n

Aholab<\/a> Taldeak eta HiTZ<\/a> zentroak euskaraz ahotsa ezagutzeko sistema berria aurkeztu dute. Aurrerapen teknologiko horrek pertsonen eta teknologiaren arteko elkarrekintza eraldatzeko ahalmena du, bereziki euskal hizkuntzaren eremuan.<\/p>\n\n\n\n

Sistema hori euskarazko 548 orduz entrenatu da, hainbat iturri publikotatik datorrena (Mozilla Common Voice 16.1<\/a>, Basque Parliament<\/a>, OpenSLR<\/a>). Horri esker, erabiltzaileek hitz egiten dituzten hitzak eta esaldiak zehaztasunez ezagutzen ditu, eta WER kalitate-maila %5etik beherakoa da.<\/p>\n\n\n\n

Aurrez aurretik entrenatutako bi eredu sortu dira, NVIDIA ereduetan oinarrituak. Horietako bat<\/a> teknika klasikoagoak erabiliz sortutako hizkuntza-eredu bat da, eta bestea<\/a>, berriz, transducer\u2011etan oinarritutako teknologia berriagoak erabiliz sortu da. Ereduen entrenamendua DIPCko zerbitzarien<\/a> Hyperion sisteman egin da.<\/p>\n\n\n\n

Garatutako sistema, potentzialki, laguntzaile birtualetan sar daiteke, mezuak bidaltzeko, informazioa bilatzeko edo agenda-notak ezartzeko. Telefono-deien erantzunak automatizatzeko aukera ere eman lezake, eraginkortasuna eta bezeroarentzako arreta hobetuz. Eta, zalantzarik gabe, euskarazko audio-grabazioen transkripzioa erraztuko du.<\/p>\n\n\n\n

Esteka honetan<\/a> dago eskuragai hizketa euskaraz ezagutzeko sistemaren demoa Sortutako ereduak ere Gaitu-Data<\/a> webgunean daude eskeragai. Ikerketa taldeak hori erabiltzeko eta teknologia hobetzen jarraitzeko iruzkinak egiteko gonbita egiten dio komunitateari. Euskal komunitatearentzat tresna baliotsua izatea eta gure hizkuntza indartzen laguntzea espero dugu.<\/p>\n","_eu_post_name":"","_eu_post_excerpt":"","_eu_post_title":"Hizketa Euskaraz Ezagutzeko Sistema eskuragarri","_en_post_content":"\n

From the Aholab<\/a> Group and the HiTZ<\/a> center we present to the public the new speech recognition system in Basque. This technological advance has the potential to transform the interaction between people and technology, especially in the field of the Basque language.<\/p>\n\n\n\n

The system has been trained with 548 hours of Basque voices from different public sources (Mozilla Common Voice 16.1<\/a>, Basque Parliament<\/a>, OpenSLR<\/a>), which allows it to accurately recognize the words and phrases spoken by users, reaching quality levels of WER less than 5%.<\/p>\n\n\n\n

Two different models have been created based on NVIDIA pre-trained models. One of them<\/a> using a language model with more classic techniques, and the other<\/a> using more emerging technologies such as transducers. The training of the models was carried out on the Hyperion system from the DIPC servers<\/a>.<\/p>\n\n\n\n

The system can potentially be integrated into virtual assistants to perform tasks such as sending messages, searching for information or setting reminders. It could also enable the automation of responses to telephone calls, improving efficiency and customer service. And it will facilitate the transcription of audio recordings in Basque.<\/p>\n\n\n\n

A demo of the speech recognition system in Basque is available at this link<\/a> and the models are available at Gaitu-Data<\/a>. The team invites the community to use it and provide feedback to continue improving the technology. We hope that it will be a valuable tool for the Basque community and contribute to the strengthening of our language.<\/p>\n","_en_post_name":"ilenia-disponible-el-sistema-de-reconocimiento-de-voz-en-euskera","_en_post_excerpt":"","_en_post_title":"Basque Speech Recognition System available","edit_language":"en","footnotes":""},"categories":[171,58,174],"tags":[],"class_list":["post-3699","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-beriak-ekitaldiak","category-noticias-y-eventos","category-noticias-y-eventos-es"],"_links":{"self":[{"href":"https:\/\/aholab.ehu.eus\/aholab\/wp-json\/wp\/v2\/posts\/3699","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=3699"}],"version-history":[{"count":8,"href":"https:\/\/aholab.ehu.eus\/aholab\/wp-json\/wp\/v2\/posts\/3699\/revisions"}],"predecessor-version":[{"id":3730,"href":"https:\/\/aholab.ehu.eus\/aholab\/wp-json\/wp\/v2\/posts\/3699\/revisions\/3730"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/aholab.ehu.eus\/aholab\/wp-json\/wp\/v2\/media\/3703"}],"wp:attachment":[{"href":"https:\/\/aholab.ehu.eus\/aholab\/wp-json\/wp\/v2\/media?parent=3699"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/aholab.ehu.eus\/aholab\/wp-json\/wp\/v2\/categories?post=3699"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/aholab.ehu.eus\/aholab\/wp-json\/wp\/v2\/tags?post=3699"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}