From 667edf9485ae6d8b44ed4825dc31789e534fc1a1 Mon Sep 17 00:00:00 2001
From: Daniel Wolf <dwolf@dannad.de>
Date: Wed, 10 Feb 2016 21:53:58 +0100
Subject: [PATCH] Improved dialog handling

---
 src/phoneExtraction.cpp | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/src/phoneExtraction.cpp b/src/phoneExtraction.cpp
index e8ccfa9..c5c9652 100644
--- a/src/phoneExtraction.cpp
+++ b/src/phoneExtraction.cpp
@@ -175,10 +175,10 @@ vector<string> extractDialogWords(string dialog) {
 	boost::algorithm::to_lower(dialog);
 
 	// Insert silences where appropriate
-	dialog = regex_replace(dialog, regex("[,;.:!?] "), " <sil> ");
+	dialog = regex_replace(dialog, regex("[,;.:!?] |-"), " <sil> ");
 
 	// Remove all undesired characters
-	dialog = regex_replace(dialog, regex("[^a-z.'\\-0-9<>]"), " ");
+	dialog = regex_replace(dialog, regex("[^a-z.'\\0-9<>]"), " ");
 
 	// Collapse whitespace
 	dialog = regex_replace(dialog, regex("\\s+"), " ");