From 4e5d2ab26cffecec4c3caee03e043ad8dcfdc49c Mon Sep 17 00:00:00 2001
From: Ziang Xie <zxie@cs.stanford.edu>
Date: Thu, 18 Jun 2015 01:39:54 -0700
Subject: [PATCH] add chars.txt and clean text scripts

---
 ctc_fast/swbd-utils/chars.txt         | 34 +++++++++++++++++++++++++++
 ctc_fast/swbd-utils/clean_text_ctc.sh | 10 ++++++++
 2 files changed, 44 insertions(+)
 create mode 100644 ctc_fast/swbd-utils/chars.txt
 create mode 100755 ctc_fast/swbd-utils/clean_text_ctc.sh

diff --git a/ctc_fast/swbd-utils/chars.txt b/ctc_fast/swbd-utils/chars.txt
new file mode 100644
index 0000000..146c0e7
--- /dev/null
+++ b/ctc_fast/swbd-utils/chars.txt
@@ -0,0 +1,34 @@
+[vocalized-noise] 1
+[laughter] 2
+' 3
+& 4
+[space] 5
+/ 6
+[noise] 7
+_ 8
+a 9
+c 10
+b 11
+e 12
+d 13
+g 14
+f 15
+i 16
+h 17
+k 18
+j 19
+m 20
+l 21
+o 22
+n 23
+q 24
+p 25
+s 26
+r 27
+u 28
+t 29
+w 30
+v 31
+y 32
+x 33
+z 34
diff --git a/ctc_fast/swbd-utils/clean_text_ctc.sh b/ctc_fast/swbd-utils/clean_text_ctc.sh
new file mode 100755
index 0000000..d64758a
--- /dev/null
+++ b/ctc_fast/swbd-utils/clean_text_ctc.sh
@@ -0,0 +1,10 @@
+#!/bin/bash
+
+for x in 'train' 'dev' 'eval2000'
+do
+    text=data/$x/text
+    ctctext=data/$x/text_ctc
+    cp $text $ctctext
+    sed -i 's/_1/ /g' $ctctext
+done
+