Update API endpoints used by Edge TTS (#412)

Fixes #411

Signed-off-by: rany <rany2@riseup.net>

Author: rany2

src/edge_tts/communicate.py

@@ -433,9 +433,9 @@ async def send_ssml_request() -> None:
             trust_env=True,
             timeout=self.session_timeout,
         ) as session, session.ws_connect(
-            f"{WSS_URL}&Sec-MS-GEC={DRM.generate_sec_ms_gec()}"
-            f"&Sec-MS-GEC-Version={SEC_MS_GEC_VERSION}"
-            f"&ConnectionId={connect_id()}",
+            f"{WSS_URL}&ConnectionId={connect_id()}"
+            f"&Sec-MS-GEC={DRM.generate_sec_ms_gec()}"
+            f"&Sec-MS-GEC-Version={SEC_MS_GEC_VERSION}",
             compress=15,
             proxy=self.proxy,
             headers=WSS_HEADERS,

src/edge_tts/constants.py

@@ -1,10 +1,14 @@
 """Constants for the edge_tts package."""
 
-BASE_URL = "speech.platform.bing.com/consumer/speech/synthesize/readaloud"
+BASE_URL = "api.msedgeservices.com/tts/cognitiveservices"
 TRUSTED_CLIENT_TOKEN = "6A5AA1D4EAFF4E9FB37E23D68491D6F4"
 
-WSS_URL = f"wss://{BASE_URL}/edge/v1?TrustedClientToken={TRUSTED_CLIENT_TOKEN}"
-VOICE_LIST = f"https://{BASE_URL}/voices/list?trustedclienttoken={TRUSTED_CLIENT_TOKEN}"
+WSS_URL = (
+    f"wss://{BASE_URL}/websocket/v1?Ocp-Apim-Subscription-Key={TRUSTED_CLIENT_TOKEN}"
+)
+VOICE_LIST = (
+    f"https://{BASE_URL}/voices/list?Ocp-Apim-Subscription-Key={TRUSTED_CLIENT_TOKEN}"
+)
 
 DEFAULT_VOICE = "en-US-EmmaMultilingualNeural"
 
@@ -22,6 +26,8 @@
     "Pragma": "no-cache",
     "Cache-Control": "no-cache",
     "Origin": "chrome-extension://jdiccldimpdaibmpdkjnbmckianbfold",
+    "Sec-WebSocket-Protocol": "synthesize",
+    "Sec-WebSocket-Version": "13",
 }
 WSS_HEADERS.update(BASE_HEADERS)
 VOICE_HEADERS = {

src/edge_tts/typing.py

@@ -20,64 +20,24 @@ class TTSChunk(TypedDict):
 class VoiceTag(TypedDict):
     """VoiceTag data."""
 
-    ContentCategories: List[
-        Literal[
-            "Cartoon",
-            "Conversation",
-            "Copilot",
-            "Dialect",
-            "General",
-            "News",
-            "Novel",
-            "Sports",
-        ]
-    ]
-    VoicePersonalities: List[
-        Literal[
-            "Approachable",
-            "Authentic",
-            "Authority",
-            "Bright",
-            "Caring",
-            "Casual",
-            "Cheerful",
-            "Clear",
-            "Comfort",
-            "Confident",
-            "Considerate",
-            "Conversational",
-            "Cute",
-            "Expressive",
-            "Friendly",
-            "Honest",
-            "Humorous",
-            "Lively",
-            "Passion",
-            "Pleasant",
-            "Positive",
-            "Professional",
-            "Rational",
-            "Reliable",
-            "Sincere",
-            "Sunshine",
-            "Warm",
-        ]
-    ]
+    ContentCategories: List[str]
+    VoicePersonalities: List[str]
 
 
 class Voice(TypedDict):
     """Voice data."""
 
     Name: str
     ShortName: str
-    Gender: Literal["Female", "Male"]
+    DisplayName: str
+    LocalName: str
+    LocaleName: str
     Locale: str
-    SuggestedCodec: Literal["audio-24khz-48kbitrate-mono-mp3"]
-    FriendlyName: str
-    Status: Literal["GA"]
+    Gender: Literal["Female", "Male"]
+    WordsPerMinute: str
+    Status: Literal["Deprecated", "GA", "Preview"]
     VoiceTag: VoiceTag
 
-
 class VoicesManagerVoice(Voice):
     """Voice data for VoicesManager."""
 

src/edge_tts/voices.py

@@ -41,17 +41,14 @@ async def __list_voices(
         data: List[Voice] = json.loads(await url.text())
 
     for voice in data:
-        # Remove leading and trailing whitespace from categories and personalities.
-        # This has only happened in one case with the zh-CN-YunjianNeural voice
-        # where there was a leading space in one of the categories.
-        voice["VoiceTag"]["ContentCategories"] = [
-            category.strip()  # type: ignore
-            for category in voice["VoiceTag"]["ContentCategories"]
-        ]
-        voice["VoiceTag"]["VoicePersonalities"] = [
-            personality.strip()  # type: ignore
-            for personality in voice["VoiceTag"]["VoicePersonalities"]
-        ]
+        if "VoiceTag" not in voice:
+            voice["VoiceTag"] = {}
+
+        if "ContentCategories" not in voice["VoiceTag"]:
+            voice["VoiceTag"]["ContentCategories"] = []
+
+        if "VoicePersonalities" not in voice["VoiceTag"]:
+            voice["VoiceTag"]["VoicePersonalities"] = []
 
     return data