Aligning the onecore code with the NVDA source onecore code

addon/synthDrivers/WorldVoice/voice/_onecore.py

@@ -46,6 +46,7 @@
 
 #: The number of 100-nanosecond units in 1 second.
 HUNDRED_NS_PER_SEC = 10000000 # 1000000000 ns per sec / 100 ns
+WAVE_HEADER_LENGTH = 46
 ocSpeech_Callback = ctypes.CFUNCTYPE(None, ctypes.c_void_p, ctypes.c_int, ctypes.c_wchar_p)
 
 class _OcSsmlConverter(speechXml.SsmlConverter):
@@ -180,9 +181,9 @@ def __init__(self, lock):
 		# Set initial values for parameters that can't be queried when prosody is not supported.
 		# This initialises our cache for the value.
 		# When prosody is supported, the values are used for cachign reasons.
-		self._rate = 50
-		self._pitch = 50
-		self._volume = 100
+		self._rate: int = 50
+		self._pitch: int = 50
+		self._volume: int = 100
 
 		if self.supportsProsodyOptions:
 			self._dll.ocSpeech_getPitch.restype = ctypes.c_double
@@ -229,7 +230,7 @@ def _maybeInitPlayer(self, wav):
 	def terminate(self):
 		# prevent any pending callbacks from interacting further with the synth.
 		self._earlyExitCB = True
-		# super().terminate()
+		super().terminate()
 		# Terminate the synth, the callback function should no longer be called after this returns.
 		self._dll.ocSpeech_terminate(self._ocSpeechToken)
 		# Drop the ctypes function instance for the callback and handle,
@@ -293,22 +294,26 @@ def pitch(self, pitch):
 			return
 		rawPitch = self._percentToParam(pitch, self.MIN_PITCH, self.MAX_PITCH)
 		self._queuedSpeech.append((self._dll.ocSpeech_setPitch, rawPitch))
+
+		# must process to apply pitch change
 		self._processQueue()
 
 	@property
-	def volume(self):
+	def volume(self) -> int:
 		if not self.supportsProsodyOptions:
 			return self._volume
 		rawVolume = self._dll.ocSpeech_getVolume(self._ocSpeechToken)
 		return int(rawVolume * 100)
 
 	@volume.setter
-	def volume(self, volume):
+	def volume(self, volume: int):
 		self._volume = volume
 		if not self.supportsProsodyOptions:
 			return
 		rawVolume = volume / 100.0
 		self._queuedSpeech.append((self._dll.ocSpeech_setVolume, rawVolume))
+
+		# must process to apply volume change
 		self._processQueue()
 
 	@property
@@ -327,6 +332,8 @@ def rate(self, rate):
 		maxRate = self.BOOSTED_MAX_RATE if self._rateBoost else self.DEFAULT_MAX_RATE
 		rawRate = self._percentToParam(rate, self.MIN_RATE, maxRate)
 		self._queuedSpeech.append((self._dll.ocSpeech_setRate, rawRate))
+
+		# must process to apply rate change
 		self._processQueue()
 
 	_rateBoost = False
@@ -419,10 +426,11 @@ def _callback(self, bytes, len, markers):
 		else:
 			self._consecutiveSpeechFailures = 0
 		# This gets called in a background thread.
-		stream = io.BytesIO(ctypes.string_at(bytes, len))
+		stream = io.BytesIO(ctypes.string_at(bytes, WAVE_HEADER_LENGTH))
 		wav = wave.open(stream, "r")
 		self._maybeInitPlayer(wav)
-		data = wav.readframes(wav.getnframes())
+		data = bytes + WAVE_HEADER_LENGTH
+		dataLen = wav.getnframes() * wav.getnchannels() * wav.getsampwidth()
 		if markers:
 			markers = markers.split('|')
 		else:
@@ -441,14 +449,17 @@ def _callback(self, bytes, len, markers):
 			# Order the equation so we don't have to do floating point.
 			pos = pos * self._bytesPerSec // HUNDRED_NS_PER_SEC
 			# Push audio up to this marker.
-			self._player.feed(data[prevPos:pos],
-				onDone=lambda index=index: synthIndexReached.notify(synth=getSynth(), index=index))
+			self._player.feed(
+				ctypes.c_void_p(data + prevPos),
+				size=pos - prevPos,
+				onDone=lambda index=index: synthIndexReached.notify(synth=getSynth(), index=index)
+			)
 			prevPos = pos
 		if self._wasCancelled:
 			if isDebugForSynthDriver():
 				log.debug("Cancelled, stopped pushing audio")
 		else:
-			self._player.feed(data[prevPos:])
+			self._player.feed(ctypes.c_void_p(data + prevPos), size=dataLen - prevPos)
 			if isDebugForSynthDriver():
 				log.debug("Done pushing audio")
 		self._processQueue()
@@ -462,8 +473,6 @@ def _getVoiceInfoFromOnecoreVoiceString(self, voiceStr):
 		language=language.replace('-','_')
 		return [ID,language,name]
 
-
-
 	@property
 	def availableVoices(self):
 		result = []
@@ -565,7 +574,6 @@ def voice(self, id):
 		voicesStr = self._dll.ocSpeech_getVoices(self._ocSpeechToken).split('|')
 		for index,voiceStr in enumerate(voicesStr):
 			ID, language, name = voiceStr.split(':')
-			language=language.replace('-','_')
 			if id == ID:
 				self._dll.ocSpeech_setVoice(self._ocSpeechToken, index)
 				return
@@ -574,7 +582,13 @@ def voice(self, id):
 	def _getDefaultVoice(self, pickAny: bool = True) -> str:
 		voicesStr = self._dll.ocSpeech_getVoices(self._ocSpeechToken).split('|')
 		for index,voiceStr in enumerate(voicesStr):
-			ID, language, name = voiceStr.split(':')
+			try:
+				print("YAA")
+				print(voiceStr)
+				ID, language, name = voiceStr.split(':')
+			except:
+				print(voiceStr)
+				raise VoiceUnsupportedError("No voices available that match the user language")
 			return ID
 		raise VoiceUnsupportedError("No voices available that match the user language")
 

buildVars.py

@@ -35,9 +35,9 @@ def _(arg):
 	# Documentation file name
 	"addon_docFileName": "readme.html",
 	# Minimum NVDA version supported (e.g. "2018.3.0", minor version is optional)
-	"addon_minimumNVDAVersion" : "2022.1",
+	"addon_minimumNVDAVersion" : "2023.3",
 	# Last NVDA version supported/tested (e.g. "2018.4.0", ideally more recent than minimum version)
-	"addon_lastTestedNVDAVersion" : "2023.1",
+	"addon_lastTestedNVDAVersion" : "2024.1",
 	# Add-on update channel (default is None, denoting stable releases,
 	# and for development releases, use "dev".)
 	# Do not change unless you know what you are doing!