TTS: Implement speech queueing on Linux and Win

5 files changed, 35 insertions, 15 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 828eb877b3..a2a09aebdc 100644
--- a/backends/text-to-speech/linux/linux-text-to-speech.cpp
+++ b/backends/text-to-speech/linux/linux-text-to-speech.cpp
@@ -128,9 +128,12 @@ Common::String LinuxTextToSpeechManager::strToUtf8(Common::String str, Common::S
 #endif
 }
 
-bool LinuxTextToSpeechManager::say(Common::String str, Common::String charset) {
+bool LinuxTextToSpeechManager::say(Common::String str, Action action, Common::String charset) {
 	if (_speechState == BROKEN)
 		return true;
+
+	if (action == DROP && isSpeaking())
+		return true;
 	
 	if (charset.empty()) {
 #ifdef USE_TRANSLATION
@@ -142,7 +145,7 @@ bool LinuxTextToSpeechManager::say(Common::String str, Common::String charset) {
 
 	str = strToUtf8(str, charset);
 
-	if (isSpeaking())
+	if (isSpeaking() && action == INTERRUPT)
 		stop();
 	if(spd_say(_connection, SPD_MESSAGE, str.c_str()) == -1) {
 		//restart the connection
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 fe7eab8ed5..49701abb4c 100644
--- a/backends/text-to-speech/linux/linux-text-to-speech.h
+++ b/backends/text-to-speech/linux/linux-text-to-speech.h
@@ -42,7 +42,7 @@ public:
 	LinuxTextToSpeechManager();
 	virtual ~LinuxTextToSpeechManager();
 
-	virtual bool say(Common::String str, Common::String charset = "");
+	virtual bool say(Common::String str, Action action, Common::String charset = "");
 
 	virtual bool stop();
 	virtual bool pause();
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 21ab0242b1..196a909f2e 100644
--- a/backends/text-to-speech/windows/windows-text-to-speech.cpp
+++ b/backends/text-to-speech/windows/windows-text-to-speech.cpp
@@ -96,12 +96,15 @@ WindowsTextToSpeechManager::~WindowsTextToSpeechManager() {
 	::CoUninitialize();
 }
 
-bool WindowsTextToSpeechManager::say(Common::String str, Common::String charset) {
-	if(_speechState == BROKEN || _speechState == NO_VOICE) {
+bool WindowsTextToSpeechManager::say(Common::String str, Action action, Common::String charset) {
+	if (_speechState == BROKEN || _speechState == NO_VOICE) {
 		warning("The tts cannot speak in this state");
 		return true;
 	}
 
+	if (isSpeaking() && action == DROP)
+		return true;
+
 	if (charset.empty()) {
 #ifdef USE_TRANSLATION
 		charset = TransMan.getCurrentCharset();
@@ -109,19 +112,18 @@ bool WindowsTextToSpeechManager::say(Common::String str, Common::String charset)
 		charset = "ASCII";
 #endif
 	}
-	if (isPaused()) {
-		resume();
-	}
-	_audio->SetState(SPAS_STOP, 0);
-	_audio->SetState(SPAS_RUN, 0);
 	// We have to set the pitch by prepending xml code at the start of the said string;
 	Common::String pitch= Common::String::format("<pitch absmiddle=\"%d\">", _ttsState->_pitch / 10);
 	str.replace((uint32)0, 0, pitch);
-
 	WCHAR *strW = Win32::ansiToUnicode(str.c_str(), Win32::getCodePageId(charset));
-	bool result = _voice->Speak(strW, SPF_ASYNC | SPF_PURGEBEFORESPEAK, NULL) != S_OK;
+
+	if ((isPaused() || isSpeaking()) && action == INTERRUPT)
+		stop();
+
+	bool result = _voice->Speak(strW, SPF_ASYNC, NULL) != S_OK;
 	free(strW);
-	_speechState = SPEAKING;
+	if (!isPaused())
+		_speechState = SPEAKING;
 	return result;
 }
 
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 f60a59de71..7a02c193ce 100644
--- a/backends/text-to-speech/windows/windows-text-to-speech.h
+++ b/backends/text-to-speech/windows/windows-text-to-speech.h
@@ -43,7 +43,7 @@ public:
 	WindowsTextToSpeechManager();
 	virtual ~WindowsTextToSpeechManager();
 
-	virtual bool say(Common::String str, Common::String charset = "");
+	virtual bool say(Common::String str, Action action, Common::String charset = "");
 
 	virtual bool stop();
 	virtual bool pause();
diff --git a/common/text-to-speech.h b/common/text-to-speech.h
index 004de3116b..b776a1cae3 100644
--- a/common/text-to-speech.h
+++ b/common/text-to-speech.h
@@ -136,6 +136,11 @@ struct TTSState {
  */
 class TextToSpeechManager {
 public:
+	enum Action {
+		INTERRUPT,
+		QUEUE,
+		DROP
+	};
 	/**
 	 * The constructor sets the language to the translation manager language if 
 	 * USE_TRANSLATION is defined, or english when it isn't defined. It sets the rate,
@@ -145,13 +150,23 @@ public:
 	virtual ~TextToSpeechManager();
 
 	/**
+	 * Interrupts what's being said and says the given string
+	 *
+	 * @param str The string to say
+	 * @param charset The encoding of the string. If empty this is assumed to be the
+	 *        encoding used for the GUI.
+	 */
+	virtual bool say(String str, String charset = "") { return say(str, INTERRUPT, charset); }
+
+	/**
 	 * Says the given string
 	 *
 	 * @param str The string to say
+	 * @param action What to do if another string is just being said.
 	 * @param charset The encoding of the string. If empty this is assumed to be the
 	 *        encoding used for the GUI.
 	 */
-	virtual bool say(String str, String charset = "") { return false; }
+	virtual bool say(String str, Action action, String charset = "") { return false; }
 
 	/**
 	 * Stops the speech