aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
Diffstat (limited to 'egs/sre16/v1/README.txt')
-rw-r--r--egs/sre16/v1/README.txt29
1 files changed, 29 insertions, 0 deletions
diff --git a/egs/sre16/v1/README.txt b/egs/sre16/v1/README.txt
new file mode 100644
index 000000000..41c00420f
--- /dev/null
+++ b/egs/sre16/v1/README.txt
@@ -0,0 +1,29 @@
1 This example demonstrates a traditional iVector system evaluated on NIST
2 SRE 2016. It is based on the recipe in ../../sre10/v1/. In addition to the
3 standard features of the SRE10 recipe, it also demonstrates the use of data
4 augmentation for PLDA training.
5
6 The recipe uses the following data for system development. This is in
7 addition to the NIST SRE 2016 dataset used for evaluation (see ../README.txt).
8
9 Corpus LDC Catalog No.
10 SWBD2 Phase 1 LDC98S75
11 SWBD2 Phase 2 LDC99S79
12 SWBD2 Phase 3 LDC2002S06
13 SWBD Cellular 1 LDC2001S13
14 SWBD Cellular 2 LDC2004S07
15 SRE2004 LDC2006S44
16 SRE2005 Train LDC2011S01
17 SRE2005 Test LDC2011S04
18 SRE2006 Train LDC2011S09
19 SRE2006 Test 1 LDC2011S10
20 SRE2006 Test 2 LDC2012S01
21 SRE2008 Train LDC2011S05
22 SRE2008 Test LDC2011S08
23 SRE2010 Eval LDC2017S06
24 Mixer 6 LDC2013S03
25
26 The following datasets are used in data augmentation.
27
28 MUSAN http://www.openslr.org/17
29 RIR_NOISES http://www.openslr.org/28