Merge branch 'master' into fix-espeak-eventloop

pyttsx3/drivers/espeak.py

@@ -12,6 +12,7 @@
 
 logger = logging.getLogger(__name__)
 
+
 if platform.system() == "Windows":
     import winsound
 
@@ -34,7 +35,12 @@ def __init__(self, proxy):
             rate = _espeak.Initialize(_espeak.AUDIO_OUTPUT_RETRIEVAL, 1000)
             if rate == -1:
                 raise RuntimeError("could not initialize espeak")
-            EspeakDriver._defaultVoice = "default"
+            current_voice = _espeak.GetCurrentVoice()
+            if current_voice and current_voice.contents.name:
+                EspeakDriver._defaultVoice = current_voice.contents.name.decode("utf-8")
+            else:
+                # Fallback to a known default if no voice is set
+                EspeakDriver._defaultVoice = "gmw/en"  # Adjust this as needed
             EspeakDriver._moduleInitialized = True
         self._proxy = proxy
         self._looping = False
@@ -77,7 +83,14 @@ def getProperty(name: str):
         if name == "voices":
             voices = []
             for v in _espeak.ListVoices(None):
-                kwargs = {"id": v.name.decode("utf-8"), "name": v.name.decode("utf-8")}
+                # Use identifier as the unique ID
+                voice_id = v.identifier.decode(
+                    "utf-8"
+                ).lower()  # Identifier corresponds to the "File" in espeak --voices
+                kwargs = {
+                    "id": voice_id,  # Use "identifier" as the ID
+                    "name": v.name.decode("utf-8"),  # Nice name
+                }
                 if v.languages:
                     try:
                         language_code_bytes = v.languages[1:]
@@ -87,14 +100,16 @@ def getProperty(name: str):
                         kwargs["languages"] = [language_code]
                     except UnicodeDecodeError:
                         kwargs["languages"] = ["Unknown"]
-                    genders = [None, "male", "female"]
+                genders = [None, "Male", "Female"]
                 kwargs["gender"] = genders[v.gender]
                 kwargs["age"] = v.age or None
                 voices.append(Voice(**kwargs))
             return voices
         if name == "voice":
             voice = _espeak.GetCurrentVoice()
-            return voice.contents.name.decode("utf-8") if voice.contents.name else None
+            if voice and voice.contents.name:
+                return voice.contents.identifier.decode("utf-8").lower()
+            return None
         if name == "rate":
             return _espeak.GetParameter(_espeak.RATE)
         if name == "volume":
@@ -110,9 +125,24 @@ def setProperty(name: str, value):
                 return
             try:
                 utf8Value = str(value).encode("utf-8")
-                _espeak.SetVoiceByName(utf8Value)
+                logging.debug(f"Attempting to set voice to: {value}")
+                result = _espeak.SetVoiceByName(utf8Value)
+                if result == 0:  # EE_OK is 0
+                    logging.debug(f"Successfully set voice to: {value}")
+                elif result == 1:  # EE_BUFFER_FULL
+                    raise ValueError(
+                        f"SetVoiceByName failed: EE_BUFFER_FULL while setting voice to {value}"
+                    )
+                elif result == 2:  # EE_INTERNAL_ERROR
+                    raise ValueError(
+                        f"SetVoiceByName failed: EE_INTERNAL_ERROR while setting voice to {value}"
+                    )
+                else:
+                    raise ValueError(
+                        f"SetVoiceByName failed with unknown return code {result} for voice: {value}"
+                    )
             except ctypes.ArgumentError as e:
-                raise ValueError(str(e))
+                raise ValueError(f"Invalid voice name: {value}, error: {e}")
         elif name == "rate":
             try:
                 _espeak.SetParameter(_espeak.RATE, value, 0)

pyttsx3/drivers/sapi5.py

@@ -15,6 +15,7 @@
 import os
 import time
 import weakref
+import locale
 
 import pythoncom
 
@@ -34,6 +35,15 @@ def buildDriver(proxy):
     return SAPI5Driver(proxy)
 
 
+def lcid_to_locale(language_code):
+    primary, sub = language_code.split("-")
+    locale_id = (int(sub) << 10) | int(primary)
+    try:
+        return locale.windows_locale[locale_id].replace("_", "-")
+    except KeyError:
+        return f"Unknown Locale: {locale_id}"
+
+
 # noinspection PyPep8Naming,PyShadowingNames
 class SAPI5Driver:
     def __init__(self, proxy):
@@ -88,7 +98,29 @@ def save_to_file(self, text, filename):
 
     @staticmethod
     def _toVoice(attr):
-        return Voice(attr.Id, attr.GetDescription())
+        # Retrieve voice ID and description (name)
+        voice_id = attr.Id
+        voice_name = attr.GetDescription()
+
+        # Retrieve and convert language code
+        language_attr = attr.GetAttribute("Language")
+        language_code = int(language_attr, 16)
+        primary_sub_code = f"{language_code & 0x3FF}-{(language_code >> 10) & 0x3FF}"
+        languages = [lcid_to_locale(primary_sub_code)]
+
+        # Retrieve gender
+        gender_attr = attr.GetAttribute("Gender")
+        gender_title_case = (gender_attr or "").title()
+        gender = gender_title_case if gender_title_case in {"Male", "Female"} else None
+
+        # Retrieve age
+        age_attr = attr.GetAttribute("Age")
+        age = age_attr if age_attr in {"Child", "Teen", "Adult", "Senior"} else None
+
+        # Create and return the Voice object with additional attributes
+        return Voice(
+            id=voice_id, name=voice_name, languages=languages, gender=gender, age=age
+        )
 
     def _tokenFromId(self, id_):
         tokens = self._tts.GetVoices()

tests/test_engines.py

@@ -41,41 +41,55 @@ def test_espeak_voices(driver_name):
     print(list(pyttsx3._activeEngines))
     print(engine)
     assert str(engine) == "espeak", "Expected engine name to be espeak"
-    voice = engine.getProperty("voice")
-    if voice:  # eSpeak-NG Windows v1.52-dev returns None
-        assert (
-            voice == "English (Great Britain)"
-        ), f"Expected {engine} default voice to be 'English (Great Britain)'"
+
+    # Verify initial voice
+    default_voice = engine.getProperty("voice")
+    print(f"Initial default voice ID: {default_voice}")
+    assert (
+        default_voice == "gmw/en"
+    ), f"Expected default voice ID to be 'gmw/en', Got: {default_voice}"
+
+    # Get and validate the number of voices
     voices = engine.getProperty("voices")
     print(f"{engine} has {len(voices) = } voices.")
     # Linux eSpeak-NG v1.50 has 109 voices,
     # macOS eSpeak-NG v1.51 has 131 voices,
     # Windows eSpeak-NG v1.52-dev has 221 voices.
-    assert len(voices) in {109, 131, 221}, f"Expected 109, 131, 221 voices in {engine}"
-    # print("\n".join(voice.id for voice in voices))
-    english_voices = [voice for voice in voices if voice.id.startswith("English")]
+    assert len(voices) in {109, 131, 221}, "Unexpected number of voices"
+
+    # Define the expected English voice IDs (excluding Caribbean for now as not in some envs
     # Linux eSpeak-NG v1.50 has 7 English voices,
     # macOS eSpeak-NG v1.51 and Windows eSpeak-NG v1.52-dev have 8 English voices.
-    assert len(english_voices) in {7, 8}, "Expected 7 or 8 English voices in {engine}"
-    names = []
-    for _voice in english_voices:
-        engine.setProperty("voice", _voice.id)
-        # English (America, New York City) --> America, New York City
-        name = _voice.id[9:-1]
-        names.append(name)
-        engine.say(f"{name} says hello")
-        engine.runAndWait()  # TODO: Remove this line when multiple utterances work!
-    name_str = "|".join(names)
-    expected = (
-        "Caribbean|Great Britain|Scotland|Lancaster|West Midlands"
-        "|Received Pronunciation|America|America, New York City"
-    )
-    no_nyc = expected.rpartition("|")[0]
-    assert name_str in {expected, no_nyc}, f"Expected '{expected}' or '{no_nyc}'."
-    print(f"({name_str.replace('|', ' ; ')})", end=" ", flush=True)
-    engine.runAndWait()
-    engine.setProperty("voice", voice)  # Reset voice to original value
-    engine.stop()
+    english_voice_ids = [
+        "gmw/en",  # Great Britain
+        "gmw/en-GB-scotland",  # Scotland
+        "gmw/en-GB-x-gbclan",  # Lancaster
+        "gmw/en-GB-x-gbcwmd",  # West Midlands
+        "gmw/en-GB-x-rp",  # Received Pronunciation
+        "gmw/en-US",  # America
+        "gmw/en-US-nyc",  # America, New York City
+    ]
+
+    for voice_id in english_voice_ids:
+        target_voice = next((v for v in voices if v.id == voice_id), None)
+        if not target_voice:
+            print(f"Voice with ID '{voice_id}' not found. Skipping.")
+            continue
+
+        print(f"Attempting to set voice to ID: {voice_id} (Name: {target_voice.name})")
+        engine.setProperty("voice", target_voice.id)
+
+        # Verify the change
+        current_voice = engine.getProperty("voice")
+        print(f"Current voice ID: {current_voice}")
+        if current_voice != target_voice.id:
+            print(
+                f"Voice change mismatch. Expected: {target_voice.id}, Got: {current_voice}. Skipping."
+            )
+            continue
+
+        engine.say(f"Hello, this is {target_voice.name}.")
+        engine.runAndWait()
 
 
 @pytest.mark.parametrize("driver_name", pyttsx3.engine.engines_by_sys_platform())

tests/test_pyttsx3.py

@@ -2,6 +2,7 @@
 
 import sys
 import wave
+from pathlib import Path
 from unittest import mock
 
 import pytest
@@ -124,10 +125,7 @@ def test_apple_nsss_voices(engine):
     engine.setProperty("voice", voice)  # Reset voice to original value
 
 
-@pytest.mark.xfail(
-    sys.platform == "darwin", reason="TODO: Fix this test to pass on macOS"
-)
-def test_saving_to_file(engine, tmp_path):
+def test_saving_to_file(engine, tmp_path: Path) -> None:
     """
     Apple writes .aiff, not .wav.  https://github.com/nateshmbhat/pyttsx3/issues/361
     """