6 files changed, 79 insertions, 18 deletions

diff --git a/backends/text-to-speech/linux/linux-text-to-speech.cpp b/backends/text-to-speech/linux/linux-text-to-speech.cpp
index 633c107727..db036e053a 100644
--- a/backends/text-to-speech/linux/linux-text-to-speech.cpp
+++ b/backends/text-to-speech/linux/linux-text-to-speech.cpp
@@ -206,7 +206,8 @@ void LinuxTextToSpeechManager::setLanguage(Common::String language) {
 void LinuxTextToSpeechManager::createVoice(int typeNumber, Common::TTSVoice::Gender gender, Common::TTSVoice::Age age, char *description) {
 	SPDVoiceType *type = (SPDVoiceType *) malloc(sizeof(SPDVoiceType));
 	*type = static_cast<SPDVoiceType>(typeNumber);
-	_ttsState->_availaibleVoices.push_back(Common::TTSVoice(gender, age, (void *) type, description));
+	Common::TTSVoice voice(gender, age, (void *) type, description);
+	_ttsState->_availaibleVoices.push_back(voice);
 }
 
 void LinuxTextToSpeechManager::updateVoices() {
@@ -238,10 +239,6 @@ bool LinuxTextToSpeechManager::popState() {
 	if (_ttsState->_next == nullptr)
 		return true;
 
-	for (Common::TTSVoice *i = _ttsState->_availaibleVoices.begin(); i < _ttsState->_availaibleVoices.end(); i++) {
-		free(i->getData());
-	}
-
 	Common::TTSState *oldState = _ttsState;
 	_ttsState = _ttsState->_next;
 
@@ -254,5 +251,9 @@ bool LinuxTextToSpeechManager::popState() {
 	return false;
 }
 
+void LinuxTextToSpeechManager::freeVoiceData(void *data) {
+	free(data);
+}
+
 
 #endif
diff --git a/backends/text-to-speech/linux/linux-text-to-speech.h b/backends/text-to-speech/linux/linux-text-to-speech.h
index 64c2371b72..57b9aeea60 100644
--- a/backends/text-to-speech/linux/linux-text-to-speech.h
+++ b/backends/text-to-speech/linux/linux-text-to-speech.h
@@ -67,6 +67,8 @@ public:
 
 	void updateState(SpeechState state);
 
+	virtual void freeVoiceData(void *data);
+
 private:
 	void init();
 	virtual void updateVoices();
diff --git a/backends/text-to-speech/windows/windows-text-to-speech.cpp b/backends/text-to-speech/windows/windows-text-to-speech.cpp
index f688ec1cd5..fec5a408d1 100644
--- a/backends/text-to-speech/windows/windows-text-to-speech.cpp
+++ b/backends/text-to-speech/windows/windows-text-to-speech.cpp
@@ -360,5 +360,9 @@ bool WindowsTextToSpeechManager::popState() {
 	setVoice(_ttsState->_activeVoice);
 	return false;
 }
+void WindowsTextToSpeechManager::freeVoiceData(void *data) {
+	ISpObjectToken *voiceToken = (ISpObjectToken *) data;
+	voiceToken->Release();
+}
 
 #endif
diff --git a/backends/text-to-speech/windows/windows-text-to-speech.h b/backends/text-to-speech/windows/windows-text-to-speech.h
index 03a1806849..2ae062ce92 100644
--- a/backends/text-to-speech/windows/windows-text-to-speech.h
+++ b/backends/text-to-speech/windows/windows-text-to-speech.h
@@ -66,6 +66,8 @@ public:
 
 	virtual bool popState();
 
+	virtual void freeVoiceData(void *data);
+
 private:
 	void init();
 	virtual void updateVoices();
diff --git a/common/text-to-speech.cpp b/common/text-to-speech.cpp
index e8e0d9b870..ce9f4a9076 100644
--- a/common/text-to-speech.cpp
+++ b/common/text-to-speech.cpp
@@ -22,9 +22,62 @@
 
 
 #include "common/text-to-speech.h"
+#include "common/system.h"
 #if defined(USE_TTS)
 
 namespace Common {
+
+TTSVoice::TTSVoice()
+	: _gender(UNKNOWN_GENDER)
+	, _age(UNKNOWN_AGE)
+	, _data(nullptr)
+	, _description("") {
+	_refCount = new int;
+	*_refCount = 1;
+}
+
+TTSVoice::TTSVoice(Gender gender, Age age, void *data, String description) 
+	: _gender(gender)
+	, _age(age)
+	, _data(data)
+	, _description(description) {
+	_refCount = new int;
+	*_refCount = 1;
+}
+
+TTSVoice::TTSVoice(const TTSVoice& voice)
+	: _gender(voice._gender)
+	, _age(voice._age)
+	, _data(voice._data)
+	, _refCount(voice._refCount)
+	, _description(voice._description) {
+	if (_data)
+		(*_refCount)++;
+}
+
+TTSVoice::~TTSVoice() {
+	// _data is a platform specific field and so it the
+	// way it is freed differs from platform to platform
+	if (--(*_refCount) == 0) {
+		if (_data)
+			g_system->getTextToSpeechManager()->freeVoiceData(_data);
+		delete _refCount;
+	}
+}
+
+TTSVoice& TTSVoice::operator=(const TTSVoice& voice) {
+	if (&voice != this) {
+		_gender = voice._gender;
+		_data = voice._data;
+		_age = voice._age;
+		_refCount = voice._refCount;
+		if (_data)
+			(*_refCount)++;
+		_description = voice._description;
+	}
+	return *this;
+}
+
 TextToSpeechManager::TextToSpeechManager() {
 	_ttsState = new TTSState;
 	_ttsState->_pitch = 0;
@@ -39,9 +92,6 @@ TextToSpeechManager::~TextToSpeechManager() {
 	TTSState *tmp = _ttsState;
 	while (tmp != nullptr) {
 		tmp = _ttsState->_next;
-		for (TTSVoice *i = _ttsState->_availaibleVoices.begin(); i < _ttsState->_availaibleVoices.end(); i++) {
-			free(i->_data);
-		}
 		delete _ttsState;
 		_ttsState = tmp;
 	}
diff --git a/common/text-to-speech.h b/common/text-to-speech.h
index 63ea1a95c0..606f8d2d4b 100644
--- a/common/text-to-speech.h
+++ b/common/text-to-speech.h
@@ -52,16 +52,15 @@ class TTSVoice {
 		};
 
 	public:
-		TTSVoice()
-			: _gender(UNKNOWN_GENDER)
-			, _age(UNKNOWN_AGE)
-			, _data(nullptr)
-			, _description("") {}
-		TTSVoice(Gender gender, Age age, void *data, String description) 
-			: _gender(gender)
-			, _age(age)
-			, _data(data)
-			, _description(description) {}
+		TTSVoice();
+
+		TTSVoice(Gender gender, Age age, void *data, String description) ;
+
+		TTSVoice(const TTSVoice& voice);
+
+		~TTSVoice();
+
+		TTSVoice& operator=(const TTSVoice& voice);
 
 		/**
 		 * Returns the gender of the used voice.
@@ -120,6 +119,7 @@ class TTSVoice {
 		Age _age;
 		void *_data;
 		String _description;
+		int *_refCount;
 };
 
 struct TTSState {
@@ -261,6 +261,8 @@ public:
 	 */
 	virtual bool popState() { return true; }
 
+	virtual void freeVoiceData(void *data) {}
+
 protected:
 	TTSState *_ttsState;