2002 NIST Speaker Recognition Evaluation
|Item Name:||2002 NIST Speaker Recognition Evaluation|
|Author(s):||Alvin Martin, Mark Przybocki|
|LDC Catalog No.:||LDC2004S04|
|Release Date:||May 21, 2004|
|Data Source(s):||broadcast news, microphone speech, telephone speech|
|Application(s):||speaker identification, speaker segmentation and tracking|
LDC User Agreement for Non-Members
|Online Documentation:||LDC2004S04 Documents|
|Licensing Instructions:||Subscription & Standard Members, and Non-Members|
|Citation:||Martin, Alvin, and Mark Przybocki. 2002 NIST Speaker Recognition Evaluation LDC2004S04. Web Download. Philadelphia: Linguistic Data Consortium, 2004.|
2002 NIST Speaker Recognition Evaluation was developed by the Linguistic Data Consortium (LDC) and the National Institute of Standards and Technology (NIST). It contains approximately 157 hours of English speech used for the NIST-sponsored 2002 Speaker Recognition Evaluation.
This corpus includes traning and test data from telephone collection projects, as well as various source data for a speaker segmentation task. The various sources are broadcast news, meeting recordings, and telephone speech. This evaluation also included the first multi-modal task, using data from an FBI voice database.
2002 NIST Speaker Recognition Evaluation is part of an ongoing series of yearly evaluations conducted by NIST. These evaluations provide an important contribution to the direction of research efforts and the calibration of technical capabilities. They are intended to be of interest to all researchers working on the general problem of text independent speaker recognition. To this end the evaluation was designed to be simple, to focus on core technology issues, to be fully supported, and to be accessible. Supporting documentation for this evaluation may be found in the 2002 NIST Speaker Recognition Evaluation Plan included in this release.
The main evaluation data was extracted from Switchboard Cellular Part 2 Audio (LDC2004S07). The extended data task used Switchboard-2 Phase II (LDC99S79) and Switchboard-2 Phase III Audio (LDC2002S06).
There are a total of 9,153 speech files (6,098 at 8 kHz and 3,055 at 16 kHz), all of which are in sphere format, for a total of about 157 hours. Here's a further breakdown:
- Training: 67 hours
- Test: 75 hours
- Various Source Segmentation Task: 15 hours
The data was initially distributed by NIST on 13 CD-ROMs (r81_1_1 through r81_13_1). This corpus consists of training and test data and replicates exactly the content and structure of the 13 CD-ROMs.
For an example of the data in this corpus, please listen to this speech sample (SPH).
None at this time.