Abstract
Introduction
Iraqi Arabic Conversational Telephone Speech was developed by Appen Pty Ltd, Sydney, Australia and contains roughly 3000 mins of speech from Iraqi Arabic speakers taking part in spontaneous telephone conversations in Colloquial Iraqi Arabic.
This corpus was collected in 2003 and 2004 by Appen Pty Ltd, Sydney, Australia. The transcripts for these calls can be found in Iraqi Arabic Conversational Telephone Speech, Transcripts (LDC2006T16).
Data
A total of 478 conversation sides from 474 unique speakers are provided, and most of these call sides comprise both sides of a conversation (that is, 202 two-channel recordings plus 74 single-channel recordings). The average duration per call is about six minutes, so each call side contains about three minutes of speech, on average.
The audio directory contains three subdirectories:
devtest -- 12 two-channel audio files
train1c -- 74 single-channel audio files
train2c -- 190 two-channel audio files
The single-channel files represent just one side of a normal conversation. The "devtest" set represents a relatively balanced (representative) sample drawn from the total pool of collected calls, based on a test-set selection process applied by the National Institute of Standards and Technology (NIST) and based on demographic, phone and audit information as provided by Appen.
All audio files are 8 kHz, 8-bit, a-law encoding, and have a NIST SPHERE formatted header.
(2006-11-17)