cond.table | R Documentation |
For averaging detection trial lists that were obtained from different
experimental conditions, the amounts of trials in these conditions
should be weighted. This function computes a contingency table from
an cst
object.
cond.table(x, cond, target = F)
x |
An object of class |
cond |
A list of factors making up the various conditions |
target |
Is column |
This function produces a table of the number of trials for each
condition in a trial list. Conditions are specified by all
combinations of all levels of the factors specified in cond
.
Conditions with no trials are left out of the table.
This table can be used in equalizing the influence of different
implicit conditions in a larger set of trials, such as NIST SRE-2008.
In order to do this properly, the analysis needs to be carried out
separately for target and non-target trials, this can be specified by
target=TRUE
.
This function is used by the functions det.sre()
and
plot.cond(), but can be used separately as well.
A data.frame with a columns for each factor, specifying the condition,
and the column Freq
that indicated the trial count.
David van Leeuwen
David A. van Leeuwen, “Overal performance metrics for multi-condition Speaker Recognition Evaluations,” Proc. Interspeech, Brighton, September 2009, pp 908–911.
plot.cond
## a NIST 2008 SRE data set
data(tno.2008)
## Analyse by gender
cond.table(tno.2008, gender)
# gender Freq
#f f 59343
#m m 39433
## Analyse by acoustic condition
cond.table(tno.2008, list(mtype, mmic, ttype, tmic))
# mtype mmic ttype tmic Freq
#1 interview mic interview mic 34181
#4 phonecall phn interview mic 7350
#8 phonecall phn phonecall mic 8454
#13 interview mic phonecall phn 11741
#16 phonecall phn phonecall phn 37050
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.