2010 NIST Speaker Recognition Evaluation Test Set 
Authors: Craig Greenberg, Alvin Martin (NIST) 
Dave Graff, Linda Brandschain, Kevin Walker (LDC)


DESCRIPTION 

The 2010 NIST Speaker Recognition Evaluation Test Set was developed by LDC
and NIST (National Institute of Standards and Technology).  The evaluation
data contains 2255 hours of American English telephone speech and speech
recorded over a microphone channel involving an interview scenario.  Some
of the data involves telephone conversations made in conditions that
involve high vocal effort on the part of the speaker.

NIST SRE is part of an ongoing series of evaluations conducted by NIST.
These evaluations are 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.

DATA SOURCES

The speech recordings in this release were collected in 2009 and 2010 by
LDC at its Human Subjects Data Collection Laboratories in
Philadelphia. This collection was part of the Mixer 6 project, which was
designed to support the development of robust speaker recognition
technology by providing carefully collected and audited speech from a large
pool of speakers recorded simultaneously across numerous microphones.

The telephone speech segments include two-channel excerpts of approximately
10 seconds and 5 minutes.  There are also summed-channel excerpts in the
range of 5 minutes.  The microphone excerpts are 3-15 minutes in
duration. As in prior evaluations, intervals of silence were not removed.
The data included in this release is 8 bit ulaw with a sample rate of 8000.

DIRECTORY STRUCTURE

In addition to evaluation data, this package also consists of answer keys,
trial and train files, development data and evaluation documentation.  The
directory structure is as follows:

/data
	/dev/NIST_SRE10_DEV_DATA
	/eval/data
	/eval/doc
	/eval/keys
	/eval/train
	/eval/trials
/docs
/README.txt (this file).

A file count of all test data found in the /data/eval/data directory is
listed by sub-directory below:

    /summed 16GB 7130 files


    /phonecall 48.9GB
        /mic 11.2 GB 2478 files
        /tel 37.7 GB 8355 files

    /interview 37.6GB
        /3min 16.2 GB 6028 files
        /8min 21.3 GB 2987 files

    /10sec 2.28GB
        /0     1.15GB 6980 files
        /1    1.13GB 6827 files



The train files in /data/eval/train provide three pieces of information for 
building models:

- model ID
- model gender
- audio ID of one or more data files in the eval data

The training conditions are described in /data/eval/doc/NIST_SRE10_evalplan.r4.pdf

The development data in /data/dev/NIST_SRE10_DEV_DATA consists of 18
interview recordings with a key for each recording and 36 phone call/speech
recordings with a key. (Each speech recording file is either a telephone
recording of the subject and interlocutor or a microphone recording of the
subject on channel "a" with the telephone channel of the interlocutor on
channel "b".  For the telephone recordings, the channel that the subject is
on is communicated in the key).

More information about the SRE10 evaluation can be found in the NIST SRE10
Evaluation Plan contained in data/eval/doc/NIST_SRE10_evalplan.r4.pdf

README created on September 14, 2016 by Karen Jones