Althingi Parliamentary Speech, Linguistic Data Consortium (LDC) Catalog Number LDC2021S01 and ISBN 1-58563-956-7, consists of approximately 542 hours of recorded speech from Althingi, the Icelandic Parliament, along with corresponding transcripts, a pronunciation dictionary and two language models. Speeches date from 2005-2016.
This dataset was collected in 2016 by the ASR for Althingi project at Reykjavik University in collaboration with the Althingi speech department. The purpose of that project was to develop an ASR (automatic speech recognition) system for parliamentary speech to replace the procedure of manually transcribing performed speeches.
The mean speech length is six minutes, with speeches ranging from under one minute to around thirty minutes. The corpus features 197 speakers (105 male, 92 female) and is split into training, development and evaluation sets. The language models are of two types: a pruned trigram model, used in decoding, and an unpruned constant ARPA 5-gram model, used for re-scoring decoding results.
Audio data is presented as single channel 16-bit mp3 files; the majority of these files have a sample rate of 44.1 kHz. Transcripts and other text data are plain text encoded in UTF-8.
Please see file.tbl for a complete file list as well as checksums for this publication.
Additional information, updates, bug fixes may be available in the LDC catalog entry for this corpus at LDC2021S01.
When publishing results based on the texts in the corpus please refer to:
Inga Rún Helgadóttir, Róbert Kjaran, Anna Björk Nikulásdóttir and Jón Guðnason, 2017. Building an ASR corpus using Althingi’s Parliamentary Speeches. Proceedings of Interspeech 2017.
Portions © 2021 Reykjavik University, © 2021 Trustees of the University of Pennsylvania