webscout 8.2.2__py3-none-any.whl → 2026.1.19__py3-none-any.whl

webscout/Provider/TTS/qwen.py

@@ -0,0 +1,334 @@
+##################################################################################
+##  Qwen3-TTS Provider                                                         ##
+##################################################################################
+import json
+import pathlib
+import random
+import string
+import tempfile
+from typing import Any, Generator, Optional, Union, cast
+
+import httpx
+from litprinter import ic
+
+from webscout import exceptions
+from webscout.litagent import LitAgent
+
+try:
+    from . import utils
+    from .base import BaseTTSProvider
+except ImportError:
+    # Handle direct execution
+    import os
+    import sys
+
+    sys.path.insert(0, os.path.join(os.path.dirname(__file__), "..", "..", ".."))
+    from webscout.Provider.TTS.base import BaseTTSProvider
+
+
+class QwenTTS(BaseTTSProvider):
+    """
+    Text-to-speech provider using the Qwen3-TTS API (Hugging Face Spaces).
+
+    This provider follows the OpenAI TTS API structure with support for:
+    - Multiple TTS models (mapped to Gradio fn_index)
+    - 40+ high-quality voices across multiple languages
+    - Automatic language detection or manual selection
+    - Multiple output formats
+    - Streaming response simulation
+    """
+
+    required_auth = False
+
+    BASE_URL = "https://qwen-qwen3-tts-demo.hf.space"
+
+    # Request headers
+    headers: dict[str, str] = {
+        "User-Agent": LitAgent().random(),
+        "origin": BASE_URL,
+        "referer": f"{BASE_URL}/",
+    }
+
+    # Override supported models
+    SUPPORTED_MODELS = ["qwen3-tts"]
+
+    # Supported voices
+    SUPPORTED_VOICES = [
+        "cherry",
+        "serena",
+        "ethan",
+        "chelsie",
+        "momo",
+        "vivian",
+        "moon",
+        "maia",
+        "kai",
+        "nofish",
+        "bella",
+        "jennifer",
+        "ryan",
+        "katerina",
+        "aiden",
+        "bodega",
+        "alek",
+        "dolce",
+        "sohee",
+        "ono_anna",
+        "lenn",
+        "sonrisa",
+        "emilien",
+        "andre",
+        "radio_gol",
+        "eldric_sage",
+        "mia",
+        "mochi",
+        "bellona",
+        "vincent",
+        "bunny",
+        "neil",
+        "elias",
+        "arthur",
+        "nini",
+        "ebona",
+        "seren",
+        "pip",
+        "stella",
+        "li",
+        "marcus",
+        "roy",
+        "peter",
+        "eric",
+        "rocky",
+        "kiki",
+        "sunny",
+        "jada",
+        "dylan",
+    ]
+
+    # Voice mapping for API compatibility
+    voice_mapping = {
+        "cherry": "Cherry / 芊悦",
+        "serena": "Serena / 苏瑶",
+        "ethan": "Ethan / 晨煦",
+        "chelsie": "Chelsie / 千雪",
+        "momo": "Momo / 茉兔",
+        "vivian": "Vivian / 十三",
+        "moon": "Moon / 月白",
+        "maia": "Maia / 四月",
+        "kai": "Kai / 凯",
+        "nofish": "Nofish / 不吃鱼",
+        "bella": "Bella / 萌宝",
+        "jennifer": "Jennifer / 詹妮弗",
+        "ryan": "Ryan / 甜茶",
+        "katerina": "Katerina / 卡捷琳娜",
+        "aiden": "Aiden / 艾登",
+        "bodega": "Bodega / 西班牙语-博德加",
+        "alek": "Alek / 俄语-阿列克",
+        "dolce": "Dolce / 意大利语-多尔切",
+        "sohee": "Sohee / 韩语-素熙",
+        "ono_anna": "Ono Anna / 日语-小野杏",
+        "lenn": "Lenn / 德语-莱恩",
+        "sonrisa": "Sonrisa / 西班牙语拉美-索尼莎",
+        "emilien": "Emilien / 法语-埃米尔安",
+        "andre": "Andre / 葡萄牙语欧-安德雷",
+        "radio_gol": "Radio Gol / 葡萄牙语巴-拉迪奥·戈尔",
+        "eldric_sage": "Eldric Sage / 精品百人-沧明子",
+        "mia": "Mia / 精品百人-乖小妹",
+        "mochi": "Mochi / 精品百人-沙小弥",
+        "bellona": "Bellona / 精品百人-燕铮莺",
+        "vincent": "Vincent / 精品百人-田叔",
+        "bunny": "Bunny / 精品百人-萌小姬",
+        "neil": "Neil / 精品百人-阿闻",
+        "elias": "Elias / 墨讲师",
+        "arthur": "Arthur / 精品百人-徐大爷",
+        "nini": "Nini / 精品百人-邻家妹妹",
+        "ebona": "Ebona / 精品百人-诡婆婆",
+        "seren": "Seren / 精品百人-小婉",
+        "pip": "Pip / 精品百人-调皮小新",
+        "stella": "Stella / 精品百人-美少女阿月",
+        "li": "Li / 南京-老李",
+        "marcus": "Marcus / 陕西-秦川",
+        "roy": "Roy / 闽南-阿杰",
+        "peter": "Peter / 天津-李彼得",
+        "eric": "Eric / 四川-程川",
+        "rocky": "Rocky / 粤语-阿强",
+        "kiki": "Kiki / 粤语-阿清",
+        "sunny": "Sunny / 四川-晴儿",
+        "jada": "Jada / 上海-阿珍",
+        "dylan": "Dylan / 北京-晓东",
+    }
+
+    def __init__(self, timeout: int = 60, proxy: Optional[str] = None):
+        """
+        Initialize the QwenTTS client.
+
+        Args:
+            timeout (int): Request timeout in seconds
+            proxy (str): Proxy configuration string
+        """
+        super().__init__()
+        self.timeout = timeout
+        self.proxy = proxy
+        self.default_voice = "cherry"
+        self.default_model = "qwen3-tts"
+
+    def _generate_session_hash(self) -> str:
+        """Generates a random session hash for Gradio."""
+        return "".join(random.choices(string.ascii_lowercase + string.digits, k=10))
+
+    def tts(self, text: str, voice: Optional[str] = None, verbose: bool = False, **kwargs) -> str:
+        """
+        Convert text to speech using Qwen3-TTS API with OpenAI-compatible parameters.
+
+        Args:
+            text (str): The text to convert to speech (max 10,000 characters)
+            **kwargs: Additional parameters (model, voice, response_format, language, verbose)
+        """
+        # Extract parameters from kwargs with defaults
+        voice = kwargs.get("voice", "cherry")
+        response_format = kwargs.get("response_format", "wav")
+        language = kwargs.get("language", "Auto / 自动")
+        verbose = kwargs.get("verbose", True)
+        if not text or not isinstance(text, str):
+            raise ValueError("Input text must be a non-empty string")
+
+        voice = self.validate_voice(voice)
+        qwen_voice = self.voice_mapping.get(voice, self.voice_mapping["cherry"])
+
+        # Create temporary file
+        file_extension = f".{response_format}"
+        filename = pathlib.Path(
+            tempfile.NamedTemporaryFile(suffix=file_extension, dir=self.temp_dir, delete=False).name
+        )
+
+        session_hash = self._generate_session_hash()
+
+        if verbose:
+            ic.configureOutput(prefix="DEBUG| ")
+            ic(f"Joining queue for voice: {voice} ({qwen_voice})")
+
+        client_kwargs: dict[str, Any] = {"headers": self.headers, "timeout": self.timeout}
+        if self.proxy:
+            client_kwargs["proxy"] = self.proxy
+
+        try:
+            with httpx.Client(**client_kwargs) as client:
+                # Step 1: Join the queue
+                join_url = f"{self.BASE_URL}/gradio_api/queue/join?"
+                payload = {
+                    "data": [text, qwen_voice, language],
+                    "event_data": None,
+                    "fn_index": 1,
+                    "trigger_id": 7,
+                    "session_hash": session_hash,
+                }
+
+                response = client.post(join_url, json=payload)
+                response.raise_for_status()
+
+                # Step 2: Poll for data (SSE)
+                data_url = f"{self.BASE_URL}/gradio_api/queue/data?session_hash={session_hash}"
+                audio_url = None
+
+                with client.stream("GET", data_url) as stream:
+                    for line in stream.iter_lines():
+                        if not line:
+                            continue
+                        if line.startswith("data: "):
+                            try:
+                                data = json.loads(line[6:])
+                            except json.JSONDecodeError:
+                                continue
+
+                            msg = data.get("msg")
+                            if msg == "process_completed":
+                                if data.get("success"):
+                                    output_data = data.get("output", {}).get("data", [])
+                                    if output_data:
+                                        audio_info = output_data[0]
+                                        path = (
+                                            audio_info["path"]
+                                            if isinstance(audio_info, dict)
+                                            else audio_info
+                                        )
+                                        audio_url = f"{self.BASE_URL}/gradio_api/file={path}"
+                                    break
+                                else:
+                                    raise exceptions.FailedToGenerateResponseError(
+                                        f"Generation failed: {data}"
+                                    )
+                            elif msg == "queue_full":
+                                raise exceptions.FailedToGenerateResponseError("Queue is full")
+
+                if not audio_url:
+                    raise exceptions.FailedToGenerateResponseError(
+                        "Failed to get audio URL from stream"
+                    )
+
+                # Step 3: Download the audio file
+                audio_response = client.get(audio_url)
+                audio_response.raise_for_status()
+
+                with open(filename, "wb") as f:
+                    f.write(audio_response.content)
+
+                if verbose:
+                    ic.configureOutput(prefix="DEBUG| ")
+                    ic(f"Speech generated successfully: {filename}")
+
+                return filename.as_posix()
+
+        except Exception as e:
+            if verbose:
+                ic.configureOutput(prefix="DEBUG| ")
+                ic(f"Error in QwenTTS: {e}")
+            raise exceptions.FailedToGenerateResponseError(f"Failed to generate audio: {e}")
+
+    def create_speech(
+        self,
+        input_text: str,
+        model: Optional[str] = "gpt-4o-mini-tts",
+        voice: Optional[str] = "alloy",
+        response_format: Optional[str] = "mp3",
+        instructions: Optional[str] = None,
+        verbose: bool = False,
+    ) -> str:
+        """
+        OpenAI-compatible speech creation interface.
+
+        Args:
+            input_text (str): The text to convert to speech
+            model (str): The TTS model to use
+            voice (str): The voice to use
+            response_format (str): Audio format
+            instructions (str): Voice instructions
+            verbose (bool): Whether to print debug information
+            **kwargs: Additional parameters
+
+        Returns:
+            str: Path to the generated audio file
+        """
+        return self.tts(
+            text=input_text,
+            voice=voice or "alloy",
+            model=model or "gpt-4o-mini-tts",
+            response_format=response_format or "mp3",
+            verbose=verbose,
+        )
+
+
+# ... (keep other classes as is)
+
+if __name__ == "__main__":
+    qwen = QwenTTS()
+    try:
+        ic.configureOutput(prefix="DEBUG| ")
+        ic("Testing Qwen3-TTS...")
+        path = qwen.create_speech(
+            input_text="Hello, this is a test.", voice="jennifer", verbose=True
+        )
+        ic.configureOutput(prefix="INFO| ")
+        ic(f"Saved to {path}")
+    except Exception as e:
+        ic.configureOutput(prefix="ERROR| ")
+        ic(f"Error: {e}")

webscout/Provider/TTS/sherpa.py

@@ -0,0 +1,286 @@
+##################################################################################
+##  SherpaTTS Provider                                                         ##
+##################################################################################
+import json
+import pathlib
+import random
+import string
+import tempfile
+from typing import Any, Generator, Optional, Union, cast
+
+import httpx
+from litprinter import ic
+
+from webscout import exceptions
+from webscout.litagent import LitAgent
+
+try:
+    from . import utils
+    from .base import BaseTTSProvider
+except ImportError:
+    # Handle direct execution
+    import os
+    import sys
+    sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..', '..', '..'))
+    from webscout.Provider.TTS.base import BaseTTSProvider
+
+class SherpaTTS(BaseTTSProvider):
+    """
+    Text-to-speech provider using the Next-gen Kaldi (Sherpa-ONNX) API.
+
+    This provider follows the OpenAI TTS API structure with support for:
+    - 50+ languages including English, Chinese, Cantonese, Arabic, French, etc.
+    - Multiple ONNX-based models (Kokoro, Piper, Coqui, etc.)
+    - Speaker ID and Speed control
+    - Multiple output formats
+    """
+    required_auth = False
+
+    BASE_URL = "https://k2-fsa-text-to-speech.hf.space"
+
+    # Request headers
+    headers: dict[str, str] = {
+        "User-Agent": LitAgent().random(),
+        "origin": BASE_URL,
+        "referer": f"{BASE_URL}/",
+    }
+
+    SUPPORTED_MODELS = [
+        "csukuangfj/kokoro-en-v0_19|11 speakers",
+        "csukuangfj/kitten-kitten-en-v0_1-fp16|8 speakers",
+        "csukuangfj/kitten-nano-en-v0_2-fp16|8 speakers",
+        "csukuangfj/kitten-nano-en-v0_1-fp16|8 speakers",
+        "csukuangfj/vits-piper-en_US-glados-high|1 speaker",
+        "csukuangfj/vits-piper-en_US-glados|1 speaker",
+        "csukuangfj/vits-piper-en_GB-southern_english_male-medium|8 speakers",
+        "csukuangfj/vits-piper-en_GB-southern_english_female-medium|6 speakers",
+        "csukuangfj/vits-piper-en_US-bryce-medium|1 speaker",
+        "csukuangfj/vits-piper-en_US-john-medium|1 speaker",
+        "csukuangfj/vits-piper-en_US-norman-medium|1 speaker",
+        "csukuangfj/vits-piper-en_US-miro-high|1 speaker",
+        "csukuangfj/vits-coqui-en-ljspeech|1 speaker",
+        "csukuangfj/vits-coqui-en-ljspeech-neon|1 speaker",
+        "csukuangfj/vits-coqui-en-vctk|109 speakers",
+        "csukuangfj/vits-piper-en_GB-miro-high|1 speaker",
+        "csukuangfj/vits-piper-en_GB-dii-high|1 speaker",
+        "csukuangfj/vits-piper-en_GB-sweetbbak-amy|1 speaker",
+        "csukuangfj/vits-piper-en_US-amy-low|1 speaker",
+        "csukuangfj/vits-piper-en_US-amy-medium|1 speaker",
+        "csukuangfj/vits-piper-en_US-arctic-medium|18 speakers",
+        "csukuangfj/vits-piper-en_US-danny-low|1 speaker",
+        "csukuangfj/vits-piper-en_US-hfc_male-medium|1 speaker",
+        "csukuangfj/vits-piper-en_US-hfc_female-medium|1 speaker",
+        "csukuangfj/vits-piper-en_US-joe-medium|1 speaker",
+        "csukuangfj/vits-piper-en_US-kathleen-low|1 speaker",
+        "csukuangfj/vits-piper-en_US-kusal-medium|1 speaker",
+        "csukuangfj/vits-piper-en_US-l2arctic-medium|24 speakers",
+        "csukuangfj/vits-piper-en_US-lessac-high|1 speaker",
+        "csukuangfj/vits-piper-en_US-lessac-low|1 speaker",
+        "csukuangfj/vits-piper-en_US-lessac-medium|1 speaker",
+        "csukuangfj/vits-piper-en_US-libritts-high|904 speakers",
+        "csukuangfj/vits-piper-en_US-libritts_r-medium|904 speakers",
+        "csukuangfj/vits-piper-en_US-ljspeech-high|1 speaker",
+        "csukuangfj/vits-piper-en_US-ljspeech-medium|1 speaker",
+        "csukuangfj/vits-piper-en_US-ryan-high|1 speaker",
+        "csukuangfj/vits-piper-en_US-ryan-low|1 speaker",
+        "csukuangfj/vits-piper-en_US-ryan-medium|1 speaker",
+        "csukuangfj/vits-piper-en_GB-alan-low|1 speaker",
+        "csukuangfj/vits-piper-en_GB-alan-medium|1 speaker",
+        "csukuangfj/vits-piper-en_GB-alan-medium",
+        "csukuangfj/vits-piper-en_GB-cori-high|1 speaker",
+        "csukuangfj/vits-piper-en_GB-cori-medium|1 speaker",
+        "csukuangfj/vits-piper-en_GB-jenny_dioco-medium|1 speaker",
+        "csukuangfj/vits-piper-en_GB-northern_english_male-medium|1 speaker",
+        "csukuangfj/vits-piper-en_GB-semaine-medium|4 speakers",
+        "csukuangfj/vits-piper-en_GB-southern_english_female-low|1 speaker",
+        "csukuangfj/vits-piper-en_GB-vctk-medium|109 speakers",
+        "csukuangfj/vits-vctk|109 speakers",
+        "csukuangfj/vits-ljs|1 speaker"
+    ]
+
+    LANGUAGES = [
+        "English", "Chinese (Mandarin, 普通话)", "Chinese+English", "Persian+English",
+        "Cantonese (粤语)", "Min-nan (闽南话)", "Arabic", "Afrikaans", "Bengali",
+        "Bulgarian", "Catalan", "Croatian", "Czech", "Danish", "Dutch", "Estonian",
+        "Finnish", "French", "Georgian", "German", "Greek", "Gujarati", "Hindi",
+        "Hungarian", "Icelandic", "Indonesian", "Irish", "Italian", "Kazakh",
+        "Korean", "Latvian", "Lithuanian", "Luxembourgish", "Maltese", "Nepali",
+        "Norwegian", "Persian", "Polish", "Portuguese", "Romanian", "Russian",
+        "Serbian", "Slovak", "Slovenian", "Spanish", "Swahili", "Swedish", "Thai",
+        "Tswana", "Turkish", "Ukrainian", "Vietnamese", "Welsh"
+    ]
+
+    def __init__(self, timeout: int = 60, proxy: Optional[str] = None):
+        """
+        Initialize the SherpaTTS client.
+        """
+        super().__init__()
+        self.timeout = timeout
+        self.proxy = proxy
+        self.default_language = "English"
+        self.default_model_choice = "csukuangfj/kokoro-en-v0_19|11 speakers"
+
+    def _generate_session_hash(self) -> str:
+        return "".join(random.choices(string.ascii_lowercase + string.digits, k=11))
+
+    def tts(self, text: str, voice: Optional[str] = None, verbose: bool = False, **kwargs) -> str:
+        """
+        Convert text to speech using Sherpa-ONNX API.
+
+        Args:
+            text: Input text
+            **kwargs: Additional parameters (language, model_choice, speaker_id, speed, response_format, verbose)
+        """
+        # Extract parameters from kwargs with defaults
+        language = kwargs.get('language', "English")
+        model_choice = kwargs.get('model_choice', "csukuangfj/kokoro-en-v0_19|11 speakers")
+        speaker_id = kwargs.get('speaker_id', "0")
+        speed = kwargs.get('speed', 1.0)
+        response_format = kwargs.get('response_format', "wav")
+        verbose = kwargs.get('verbose', True)
+        if not text:
+            raise ValueError("Input text must be a non-empty string")
+
+        model_choice = self.validate_model(model_choice)
+
+        session_hash = self._generate_session_hash()
+        filename = pathlib.Path(tempfile.NamedTemporaryFile(suffix=f".{response_format}", dir=self.temp_dir, delete=False).name)
+
+        if verbose:
+            ic.configureOutput(prefix='DEBUG| ')
+            ic(f"SherpaTTS: Generating speech for '{text[:20]}...' using {language}/{model_choice}")
+
+        client_kwargs: dict[str, Any] = {"headers": self.headers, "timeout": self.timeout}
+        if self.proxy:
+            client_kwargs["proxy"] = self.proxy
+
+        try:
+            with httpx.Client(**client_kwargs) as client:
+                # Step 1: Join the queue
+                join_url = f"{self.BASE_URL}/gradio_api/queue/join?"
+                payload = {
+                    "data": [language, model_choice, text, speaker_id, speed],
+                    "event_data": None,
+                    "fn_index": 1,
+                    "trigger_id": 9,
+                    "session_hash": session_hash
+                }
+
+                response = client.post(join_url, json=payload)
+                response.raise_for_status()
+
+                # Step 2: Poll for data
+                data_url = f"{self.BASE_URL}/gradio_api/queue/data?session_hash={session_hash}"
+                audio_url = None
+
+                with client.stream("GET", data_url) as stream:
+                    for line in stream.iter_lines():
+                        if not line:
+                            continue
+                        if line.startswith("data: "):
+                            try:
+                                data = json.loads(line[6:])
+                            except json.JSONDecodeError:
+                                continue
+
+                            msg = data.get("msg")
+                            if msg == "process_completed":
+                                if data.get("success"):
+                                    output_data = data.get("output", {}).get("data", [])
+                                    if output_data:
+                                        audio_info = output_data[0]
+                                        path = audio_info["path"] if isinstance(audio_info, dict) else audio_info
+                                        audio_url = f"{self.BASE_URL}/gradio_api/file={path}"
+                                    break
+                                else:
+                                    raise exceptions.FailedToGenerateResponseError(f"Generation failed: {data}")
+                            elif msg == "queue_full":
+                                raise exceptions.FailedToGenerateResponseError("Queue is full")
+
+                if not audio_url:
+                    raise exceptions.FailedToGenerateResponseError("Failed to get audio URL from stream")
+
+                # Step 3: Download the audio file
+                audio_response = client.get(audio_url)
+                audio_response.raise_for_status()
+
+                with open(filename, "wb") as f:
+                    f.write(audio_response.content)
+
+                if verbose:
+                    ic.configureOutput(prefix='DEBUG| ')
+                    ic(f"Speech generated successfully: {filename}")
+
+                return filename.as_posix()
+
+        except Exception as e:
+            if verbose:
+                ic.configureOutput(prefix='DEBUG| ')
+                ic(f"Error in SherpaTTS: {e}")
+            raise exceptions.FailedToGenerateResponseError(f"Failed to generate audio: {e}")
+
+    def create_speech(
+        self,
+        input_text: str,
+        model: Optional[str] = "csukuangfj/kokoro-en-v0_19|11 speakers",
+        voice: Optional[str] = None,
+        response_format: Optional[str] = "mp3",
+        instructions: Optional[str] = None,
+        verbose: bool = False
+    ) -> str:
+        """
+        OpenAI-compatible speech creation interface.
+
+        Args:
+            input_text (str): The text to convert to speech
+            model (str): The TTS model to use
+            voice (str): The voice to use (not used by SherpaAI directly)
+            response_format (str): Audio format
+            instructions (str): Voice instructions
+            verbose (bool): Whether to print debug information
+
+        Returns:
+            str: Path to the generated audio file
+        """
+        model_choice = model or "csukuangfj/kokoro-en-v0_19|11 speakers"
+        return self.tts(
+            text=input_text,
+            model_choice=model_choice,
+            response_format=response_format or "mp3",
+            verbose=verbose
+        )
+
+    def with_streaming_response(self):
+        return StreamingResponseContextManager(self)
+
+class StreamingResponseContextManager:
+    def __init__(self, tts_provider: SherpaTTS):
+        self.tts_provider = tts_provider
+    def create(self, **kwargs):
+        audio_file = self.tts_provider.create_speech(**kwargs)
+        return StreamingResponse(audio_file)
+    def __enter__(self): return self
+    def __exit__(self, exc_type, exc_val, exc_tb): pass
+
+class StreamingResponse:
+    def __init__(self, audio_file: str):
+        self.audio_file = audio_file
+    def __enter__(self): return self
+    def __exit__(self, exc_type, exc_val, exc_tb): pass
+    def stream_to_file(self, file_path: str):
+        import shutil
+        shutil.copy2(self.audio_file, file_path)
+    def iter_bytes(self, chunk_size: int = 1024):
+        with open(self.audio_file, 'rb') as f:
+            while chunk := f.read(chunk_size):
+                yield chunk
+
+if __name__ == "__main__":
+    tts = SherpaTTS()
+    try:
+        path = tts.tts("This is a Sherpa-ONNX test.", verbose=True)
+        ic.configureOutput(prefix='INFO| ')
+        ic(f"Result: {path}")
+    except Exception as e:
+        ic.configureOutput(prefix='ERROR| ')
+        ic(f"Error: {e}")