HiTZ GitHub plataforman HiTZ zentroko ikertzaileek garatutako softwarea, tresnak eta ikerketa-proiektuen kode irekia aurki daitezke. Hizkuntza eta hizketa teknologien arloko baliabideak eskuratu, eta gure garapenak gertutik ezagutu.
Arakatu gure kode irekiko proiektuak GitHuben.
Esteka: https://github.com/hitz-zentroa
HiTZ Zenodo komunitatean HiTZ zentroko ikertzaileek sortutako argitalpenak, datu-multzoak, softwarea eta bestelako ikerketa-baliabide irekiak aurki daitezke. Ezagutu gure ikerketaren emaitzak eta eskuratu komunitate zientifikoarekin partekatutako materialak.
Arakatu gure argitalpen, datu-multzo eta software irekiak Zenodon.
Esteka: https://zenodo.org/communities/hitz/
HiTZ Hugging Face atarian HiTZ zentroak garatutako hizkuntza eta hizketa teknologietako baliabide irekiak aurki daitezke, besteak beste LLM, ASR eta TTS ereduak. Ezagutu gure ikerketaren emaitzak eta deskargatu komunitatearekin partekatutako ereduak eta datu-multzoak.
Ezagutu gure eredu eta datu-multzo irekiak Hugging Face plataforman.
Esteka: https://huggingface.co/HiTZ
AhoTTS Aholabek garatutako testutik hizketarako sintetizadorea da, euskararako eta gaztelaniarako. GitHub biltegian kodea, prozesamendu linguistikoa eta ahotsak sortzeko baliabideak daude eskuragarri.
Esteka: https://github.com/aholab/AhoTTS
Basadas en distintas configuraciones sintácticas, las frases semánticamente impredecibles están en formato texto.
Descarga: https://aholab.ehu.eus/aholab/download/100-frases-semanticamente-impredecibles/
A reference SAMPA code page for Basque Language, available at
Bizkaifon: speech and video database for the Western dialects of the Basque Language
Bizkaifon is a multimodal (speech and video) database containing thousands recordings of the many different western dialects of Basque. Most of them are transcribed to Standard Basque. It is accessible via web through this page. The database is available at ELRA.
500 phonetically balanced sentences recorded by Male (Kepa) and Female (Jaione) voice talents for 4 emotions (Neutral, Joy, Sadness & Anger) . It was registered at 48kHz, 16bits, semi-professional room, 2 microphones and laryngograph included.
Kattalin and Peru
702 phonetically balanced sentences repeated for the 6 MPEG emotions and neutral style, for male (Peru) and female (Kattalin) voices. It also contains a continuous read speech of 8 min, also in 7 styles. It was registered at 48kHz, 16bits, semi-professional room, 2 microphones and laryngograph included. The female voice Kattalin has been segmented at phone level and manually revised for the neutral style. It is available for research use. Please send us an email.
170 items (sentences, words) repeated for the 6 MPEG emotions and neutral style, for a female voice. The audio was registered at 32kHz, 16bits, professional studio, 1 microphones and laryngograph included. Two cameras were used for video recordings (one frontal, one lateral). Face markers were used. The female voice has been segmented at phone level and manually revised for the neutral style. It is available for research use. Please send us an email.
Navas, E., Hernáez, I., Castelruiz, A., Luengo, I. Obtaining and Evaluating an Emotional Database for Prosody Modelling in Standard Basque. LNAI 3206, pp. 393–400, 2004. ISSN: 0302-9743
PDF
500 phonetically balanced sentences recorded by 4 Male and 3 Female voice talents for Neutral style . It was registered at 48kHz, 16bits, semi-professional room, 2 microphones and laryngograph included.
Described in Luengo, I., Saratxaga, I., Navas, E., Hernáez, I., Sanchez, J., Sainz, I. Evaluation Of Pitch Detection Algorithms Under Real Conditions. Proc. of 32nd IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), pp. 1057-1060, Honolulu, 2007. ISBN: 1-4244-0728-1 PDF
Please send us an email
AhoVAD: Voice Detection Algorithm
Described in
Luengo, I., Navas, E., Odriozola, I., Saratxaga, I., Hernáez, I., Sainz, I., Erro, D. Modified LTSE-VAD Algorithm for Applications Requiring Reduced Silence Frame Misclassification.
Proc. of the Seventh International Language Resources and Evaluation (LREC’10), paper 741, pp. 1539-1544, 2010. ISBN: 2-9517408-6-7. PDF
Please send us an email
3799 (Basque) and 3995 (Spanish) phonetically balanced sentences recorded by Male and Female voice talents in neutral style.
It was registered at 48kHz, 16bits, semi-professional room, 2 microphones and laryngograph included.
This database contains 12 sentences written in standard Basque and uttered by 9 male speakers at slow, normal and fast rate, each repeated 7 times (total of 2268 sentences).
It was developed under the interest of Dr. Fujisaki after his short visit to our laboratories.
It is available for research use. Please send us an email.
The Basque MDB-600 database contains the recordings of 660 speakers of Basque recorded over the mobile telephone network.
This database is partitioned into 4 CDs
The database complies with the common specifications created in the SpeechDat project.
