This is the new version of the ELRA Catalogue of Language Resources. If you would like to view the older version, click here

An increasing number of LRs in the various fields of Human Language Technology (see image on the left-hand side) are distributed on behalf of ELRA via its operational body ELDA, thanks to the contribution of various players of the HLT community.
Our aim is to provide Language Resources, by means of this repository, so as to prevent researchers and developers from investing efforts to rebuild resources which already exist as well as help them identify and access those resources.

Latest Resources

• GlobalPhone Multilingual Model Package

  The GlobalPhone Multilingual Model Package contains about 22 hours of transcribed read speech spoken by native speakers in 22 languages. The data are sampled from the GlobalPhone Speech and Text Data available in the ELRA Catalogue, i.e.: Arabic (ELRA-S0192), Bulgarian (ELRA-S0319), Chinese-Mandarin (ELRA-S0193), Chinese-Shanghai (ELRA-S0194), Croatian (ELRA-S0195), Czech (ELRA-S0196), French ...

• Normalized Arabic Fragments for Inestimable Stemming (NAFIS)

  Normalized Arabic Fragments for Inestimable Stemming (NAFIS) is an Arabic stemming gold standard corpus composed by a collection of sentences, selected to be representative of Arabic stemming tasks and manually annotated. Indeed, NAFIS is: Comprehensive: The content of NAFIS can be generalized to the Arabic language as a whole. Within ...

• Persian Audio Dictionary

  This dictionary consists of more than 50,000 entries (along with almost all wordforms and proper names) with corresponding audio files in MP3 and English transliterations. The words have been recorded with standard Persian (Farsi) pronunciation (all by a single speaker). This dictionary is provided with a software produced by Persian ...

• GlobalPhone 2000 Speaker Package

  The GlobalPhone 2000 Speaker Package contains transcribed read speech spoken by 2000 native speakers in 22 languages. The data are sampled from the GlobalPhone Speech and Text Data available in the ELRA Catalogue, i.e.: Arabic (ELRA-S0192), Bulgarian (ELRA-S0319), Chinese-Mandarin (ELRA-S0193), Chinese-Shanghai (ELRA-S0194), Croatian (ELRA-S0195), Czech (ELRA-S0196), French (ELRA-S0197), German (ELRA-S0198), ...

• Training and test data for Arabizi detection and transliteration

  The dataset is composed of two distinct resources: 1) A collection of mixed English and Arabizi text intended to train and test a system for the automatic detection of code-switching in mixed English and Arabizi texts. The training part of the corpus contains: 522 tweets composed of 5,207 tokens (including ...

• Mbochi speech corpus

  The Mbochi speech corpus was developed in the framework of ANR-DFG BULB project. This project aims to provide field linguists (eg working on morphology) with tools for less or not written languages. The provided corpus is a subset from the corpus developed in this framework. The provided corpus consists of ...

• Chinese Mandarin (South) database

  This database contains the recordings of 1000 Chinese Mandarin speakers from Southern China (500 males and 500 females), from 18 to 60 years’ old, recorded in quiet studios located in Shenzhen and in Hong Kong Special Administrative Region, People’s Republic of China. Demographics of native speakers from Southern China is ...

• Chinese Mandarin (North) database

  This database contains the recordings of 500 Chinese Mandarin speakers from Northern China (250 males and 250 females), from 18 to 60 years’ old, recorded in quiet studios located in Shenzhen and in Hong Kong Special Administrative Region, People’s Republic of China. Demographics of native speakers from Northern China is ...

• TRAD Chinese-French Parallel Text - Blog

  TRAD Chinese-French Parallel Text - Blog was developed by ELDA as part of the PEA-TRAD project. It contains English translations of a subset of approximately 10,000 Chinese words from GALE Phase 1 Chinese Blog Parallel Text (available here: https://catalog.ldc.upenn.edu/LDC2008T06). The PEA-TRAD project (Translation as a Support for Document Analysis) was ...

• English-Vietnamese Parallel Corpus

  This is a corpus of 500,000 English-Vietnamese sentence pairs, built to develop SMT (Statistical Machine Translation) systems. The parallel corpus contains English documents translated by professional translators into Vietnamese. The source texts include books, dictionaries, newspapers, online news, collected between 2000 and 2007. All Vietnamese sentences have been word-segmented and ...

• Nautilus Speaker Characterization (NSC) Corpus

  The Nautilus Speaker Characterization (NSC) Corpus comprises clean microphone recordings of conversational speech from 300 German speakers (126 males and 174 females) aged 18 to 35 years, with no marked dialect/accent. The recordings were performed in the acoustically-isolated room "Nautilus" (which gives name to this database) of the Quality and ...

• French dictionary of definitions (SYNAPSE)

  The French dictionary of definitions (SYNAPSE) was designed by linguists graduated in France with the objective of being used on digital media. Consequently: - definitions are provided on a linear format with a classification based on syntactic form (and not under a hierarchic form of senses), - definitions of each ...

• Metalogue Multi-Issue Bargaining Dialogue

  INTRODUCTION Metalogue Multi-Issue Bargaining Dialogue was developed by the Metalogue Consortium (http://cordis.europa.eu/project/rcn/110655_en.html) under the European Community's Seventh Framework Programme for Research and Technological Development (https://ec.europa.eu/research/fp7/index_en.cfm). This release consists of approximately 2.5 hours of semantically annotated English dialogue data that includes speech and transcripts. The goal of the Metalogue project was ...

• 2007 CoNLL Shared Task - Basque, Catalan, Czech & Turkish

  2007 CoNLL Shared Task - Basque, Catalan, Czech & Turkish consists of dependency treebanks in four languages used as part of the CoNLL 2007 shared task on multi-lingual dependency parsing and domain adaptation. The languages covered in this release are: Basque, Catalan, Czech and Turkish. The Conference on Computational Natural ...

• 2007 CoNLL Shared Task - Greek, Hungarian & Italian

  2007 CoNLL Shared Task - Greek, Hungarian & Italian consists of dependency treebanks in three languages used as part of the CoNLL 2007 shared task on multi-lingual dependency parsing and domain adaptation. The languages covered in this release are: Greek, Hungarian and Italian. The Conference on Computational Natural Language Learning ...

• 2007 CoNLL Shared Task - Arabic & English

  2007 CoNLL Shared Task - Arabic & English consists of dependency treebanks in two languages used as part of the CoNLL 2007 shared task on multi-lingual dependency parsing and domain adaptation. The languages covered in this release are: Arabic and English. The Conference on Computational Natural Language Learning (CoNLL) is ...

• Arabic dictionary of inflected words with recognition of agglutinated clitics and inflection system

  This dictionary consists of 6 million inflected forms, fully vowelized, generated in compliance with the grammatical rules of Arabic and tagged with grammatical information which includes POS and grammatical features, including number, gender, case, definiteness, tense, mood and compatibility with clitic agglutination. It is accompanied by a grammatical resource that ...

• Arabic dictionary of inflected words

  The Arabic dictionary of inflected words consists of a list of 6 million inflected forms, fully vowelized, generated in compliance with the grammatical rules of Arabic and tagged with grammatical information which includes POS and grammatical features, including number, gender, case, definiteness, tense, mood and compatibility with clitic agglutination. The ...

• Helsinki Corpus of Swahili

  This is a text corpus of Swahili language of 25 million words, annotated for part-of-speech, morphology and syntax. The corpus contains prose text from fiction, news media and government documents domains, from the period between 1953 and 2016. This package contains: - the Helsinki Corpus of Swahili 2.0 Non Annotated ...

• NUM 5M Mongolian written corpus

  This is a corpus of Mongolian text mostly from domains like online or printed daily newspapers, literature, and laws. The collected raw texts was reduced from 5 to 4.8 million words after cleaning. The cleaned corpus comprises: - 144 texts from laws until 2009, - 288 texts from literature that ...

• Persian Speech Corpus

  This about 2.5-hour Single-Speaker Speech corpus has been developed using the same methodologies used in the PhD work carried out by Nawar Halabi at the University of Southampton. The corpus was recorded in Persian (Tehrani accent) by one male speaker using a professional studio, through a "Blubbery" model microphone of ...

• English-Persian parallel corpus

  The English-Persian parallel corpus contains more than 200,000 aligned sentences across a variety of text types from the domains of art, law, culture, science, religion, literature, medicine, idioms, politics and others. It is an extension of the English-Persian parallel corpus already distributed by ELRA (Catalogue Reference: ELRA-W0051). This new version ...

• Pashto phonetic lexicon

  This is a phonetic lexicon of 21,560 tokens in Pashto with their phonetic transcription in IPA. It covers the major dialect of the TRAD Pashto Broadcast News Speech Corpus (see ELRA Catalogue reference ELRA-S0381) from which the most frequent words were extracted. The pronunciation dictionary of these words was manually ...

• The FAME! Speech Corpus

  The components of the Frisian data collection are speech and language resources gathered for building a large vocabulary ASR system for the Frisian language. Firstly, a new broadcast database is created by collecting recordings from the archives of the regional broadcaster Omrop Fryslân, and annotating them with various information such ...

• GlobalPhone Bulgarian Pronunciation Dictionary 260k entries (extended version)

  This extended version of the Bulgarian Pronunciation Dictionary called Bulgarian-Dict260k contains pronunciations of more than 260,000 word forms. The dictionary matches in phone set and format the original GlobalPhone Bulgarian Pronunciation Dictionary (see ELRA-S0351) of 20,000 word forms. Bulgarian-Dict260k was built based on the extension of the Bulgarian GlobalPhone text ...

• Danish Propbank

  The Danish Propbank (DPB) is a multi-layer treebank, annotated not only with morphosyntactic, but also with semantic information, in particular propositions/frames with VerbNet classes and semantic roles for both arguments and satellites. In addition, the corpus has been annotated with 20 Named Entity classes and a 200-category semantic ontology for ...

• Accented English GlobalPhone

  The Accented English part of the GlobalPhone resources contains 63 recording sessions of Bulgarian, Chinese, German, and Indian native speakers reading 37 English sentences each, produced in GlobalPhone-style, i.e. 16kHz PCM encoded audio recordings of utterance-segmented read speech from the newspaper domain.

• Parallel EMG-Acoustic English GlobalPhone

  The parallel EMG-Acoustic English GlobalPhone language resource contains 63 recordings sessions from 8 speakers articulating speech in three speaking modes, audible, whispered, and silent by reading three times 50 English sentences in GlobalPhone-style, i.e. 16kHz PCM encoded audio recordings of utterance-segmented read speech from the newspaper domain. Speech is recorded ...

• ETAPE Evaluation Package

  The ETAPE project (Evaluation en Traitement Automatique de la Parole) consists in an evaluation campaign for automatic speech processing systems. The project was funded by the French National Research Agency (ANR) under grant agreement ANR-09-CORD-009. The ETAPE 2011 campaign follows the series of ESTER campaigns organized in 2003, 2005 and ...

• SALA II US English database (2000 speakers)

  The SALA II US English database collected in the United States was recorded within the scope of the SALA II project. It contains the recordings of ca. 2,000 US English speakers (equally balanced between males and females, including some speakers with Hispanic accents) recorded over the United States mobile telephone ...

• TRAD Chinese-French Email Parallel corpus – Development Set

  This is a parallel corpus of 15,000 characters in Chinese (equivalent to 10,000 words) and a reference translation in French. The source texts are a selection of emails from the Speechocean King-NLP-001 corpus, a corpus of private emails collected from the daily life and business domains. The content has also ...

• TRAD Chinese-English Email Parallel corpus – Development Set

  This is a parallel corpus of 15,000 characters in Chinese (equivalent to 10,000 words) and a reference translation in English. The source texts are a selection of emails from the Speechocean King-NLP-001 corpus, a corpus of private emails collected from the daily life and business domains. The content has also ...

• TRAD Chinese-English Email Parallel corpus – Test Set

  This is a parallel corpus of 15,000 characters in Chinese (equivalent to 10,000 words) and 2 reference translations in English. The source texts are a selection of emails from the Speechocean King-NLP-001 corpus, a corpus of private emails collected from the daily life and business domains. The translation has been ...

• TRAD Chinese-French Email Parallel corpus – Test Set

  This is a parallel corpus of 15,000 characters in Chinese (equivalent to 10,000 words) and 2 reference translations in French. The source texts are a selection of emails from the Speechocean King-NLP-001 corpus, a corpus of private emails collected from the daily life and business domains. The translation has been ...

• SecuVoice

  SecuVoice is a corpus of single-channel utterances in Spanish containing sequences of isolated digits from zero to nine. These utterances were acquired by using two different devices, i.e. a mid-range smartphone and a high-range one. For both models, the utterances were stored as uncompressed monophonic WAV files with a sampling ...

• TRAD Arabic-English Newspaper Parallel corpus - Test set 1

  This is a parallel corpus of 10,000 words in Arabic and 2 reference translations in English. The source texts are articles collected in 2012 from the Arabic version of Le Monde Diplomatique. The translation has been conducted by two different translation teams following a strict protocol aimed at producing high ...

• TRAD Chinese-French News Articles Parallel corpus

  This is a parallel corpus of 15,000 characters in Chinese (equivalent to 10,000 words) and 2 reference translations in French. The source texts are newspaper articles from the Chinese version of Voice of America. Articles are dated from 2011 and 2012. The translation has been conducted by two different translation ...

• TRAD Arabic-French Mailing lists Parallel corpus - Test set

  This is a parallel corpus of 10,000 words in Arabic and 4 reference translations in French. The source texts are emails collected from Wikiar-I, a mailing list for discussions about the Arabic Wikipedia. The collected emails are dated from 2010 to 2012. The translation has been conducted by four different ...

• TRAD Arabic-French Newspaper Parallel corpus - Test set 2

  This is a parallel corpus of 10,000 words in Arabic and 2 reference translations in French. The source texts are articles collected in May 2013 from the Arabic version of Le Monde Diplomatique. The translation has been conducted by two different translation teams following a strict protocol aimed at producing ...

• TRAD Chinese-French Web domain (blogs) Parallel corpus

  This is a parallel corpus of 15,000 characters in Chinese (equivalent to 10,000 words) and 2 reference translations in French. The source texts are blog articles dealing with various subjects such as economy, environment, society, technologies, etc. Articles are dated from June 2013. The translation has been conducted by two ...

• TRAD Chinese-English Web domain (blogs) Parallel corpus

  This is a parallel corpus of 15,000 characters in Chinese (equivalent to 10,000 words) and 2 reference translations in English. The source texts are blog articles dealing with various subjects such as economy, environment, society, technologies, etc. Articles are dated from June 2013. The translation has been conducted by two ...

• TRAD Chinese-English News Articles Parallel corpus

  This is a parallel corpus of 15,000 characters in Chinese (equivalent to 10,000 words) and 2 reference translations in English. The source texts are newspaper articles from the Chinese version of Voice of America. Articles are dated from 2011 and 2012. The translation has been conducted by two different translation ...

• TRAD Arabic-French Newspaper Parallel corpus - Test set 1

  This is a parallel corpus of 10,000 words in Arabic and 4 reference translations in French. The source texts are articles from the Arabic version of Le Monde Diplomatique collected in 2012. The translation has been conducted by four different translation teams following a strict protocol aimed at producing high ...

• TRAD Arabic-English Web domain (blogs) Parallel corpus

  This is a parallel corpus of 10,000 words in Arabic and 2 reference translations in English. The source texts are blog articles written between 2008 and 2013. The translation has been conducted by two different translation teams following a strict protocol aimed at producing high quality translations. The content is ...

• TRAD Arabic-English Mailing lists Parallel corpus - Test set

  This is a parallel corpus of 10,000 words in Arabic and 2 reference translations in English. The source texts are emails collected from Wikiar-I, a mailing list for discussions about the Arabic Wikipedia. The collected emails are dated from 2010 to 2012. The translation has been conducted by two different ...

• TRAD Arabic-French Web domain (blogs) Parallel corpus

  This is a parallel corpus of 10,000 words in Arabic and 4 reference translations in French. The source texts are blog articles from 2008 to 2013. The translation has been conducted by four different translation teams following a strict protocol aimed at producing high quality translations. The content is also ...

• TRAD Arabic-French Mailing lists Parallel corpus - Development set

  This is a parallel corpus of 10,000 words in Arabic and a reference translation in French. The source texts are emails collected from Wikiar-I, a mailing list for discussions about the Arabic Wikipedia. The collected emails are dated from 2004 to 2007. The translation has been conducted following a strict ...

• TRAD Arabic-English Parallel corpus of transcribed Broadcast News Speech

  This is a parallel corpus of 10,000 words in Arabic and 2 reference translations in English. The source texts are transcriptions of broadcast news in Arabic recorded on France 24. The translation has been conducted by two different translation teams following a strict protocol aimed at producing high quality translations. ...

• TRAD Arabic-English Mailing lists Parallel corpus - Development set

  This is a parallel corpus of 10,000 words in Arabic and a reference translation in English. The source texts are emails collected from Wikiar-I, a mailing list for discussions about the Arabic Wikipedia. The collected emails are dated from 2004 to 2007. The translation has been conducted following a strict ...

• TRAD Arabic-French Parallel corpus of transcribed Broadcast News Speech

  This is a parallel corpus of 10,000 words in Arabic and 4 reference translations in French. The source texts are transcriptions of broadcast news in Arabic recorded on France 24. The translation has been conducted by four different translation teams following a strict protocol aimed at producing high quality translations. ...

• Serbian emotional speech database (GEES)

  The database contains recordings from six actors, three of each gender. The following emotions have been recorded: neutral, anger, happiness, sadness and fear. The database consists of: 32 isolated words, 30 short semantically neutral sentences, 30 long semantically neutral sentences and one passage with 79 words in size. The overall ...

• Arabic Speech Corpus

  This speech corpus has been developed as part of a PhD work carried out by Nawar Halabi at the University of Southampton. The corpus was recorded through a Neumann TLM 103 Studio Microphone by one male speaker in South Levantine Arabic (Damascian accent) in a professional studio. The transcript was ...

• Collins Multilingual database (MLD) – WordBank with audio files

  The Collins Multilingual database covers Real Life Daily vocabulary. It is composed of a multilingual lexicon in 32 languages (the WordBank, see ELRA-T0376) and a multilingual set of sentences in 28 languages (the PhraseBank, see ELRA-T0377). This version includes the corresponding audio files covering 26 languages of the 32 languages ...

• Collins Multilingual database (MLD) – PhraseBank with audio files

  The Collins Multilingual database covers Real Life Daily vocabulary. It is composed of a multilingual lexicon in 32 languages (the WordBank, see ELRA-T0376) and a multilingual set of sentences in 28 languages (the PhraseBank, see ELRA-T0377). This version includes the audio files corresponding to each phrase in the Collins MLD ...