feat(tools): add Typecast tool integration with text to speech features

libs/community/langchain_community/tools/__init__.py

@@ -317,6 +317,9 @@
         TavilyAnswer,
         TavilySearchResults,
     )
+    from langchain_community.tools.typecast.text2speech import (
+        TypecastText2SpeechTool,
+    )
     from langchain_community.tools.vectorstore.tool import (
         VectorStoreQATool,
         VectorStoreQAWithSourcesTool,
@@ -402,6 +405,7 @@
     "EdenAiTextToSpeechTool",
     "EdenaiTool",
     "ElevenLabsText2SpeechTool",
+    "TypecastText2SpeechTool",
     "ExtractHyperlinksTool",
     "ExtractTextTool",
     "FileSearchTool",
@@ -556,6 +560,7 @@
     "EdenAiTextToSpeechTool": "langchain_community.tools.edenai",
     "EdenaiTool": "langchain_community.tools.edenai",
     "ElevenLabsText2SpeechTool": "langchain_community.tools.eleven_labs.text2speech",
+    "TypecastText2SpeechTool": "langchain_community.tools.typecast.text2speech",
     "ExtractHyperlinksTool": "langchain_community.tools.playwright",
     "ExtractTextTool": "langchain_community.tools.playwright",
     "FileSearchTool": "langchain_community.tools.file_management",

libs/community/langchain_community/tools/typecast/__init__.py

@@ -0,0 +1,5 @@
+"""Typecast Services Tools."""
+
+from langchain_community.tools.typecast.text2speech import TypecastText2SpeechTool
+
+__all__ = ["TypecastText2SpeechTool"]

libs/community/langchain_community/tools/typecast/text2speech.py

@@ -0,0 +1,118 @@
+import tempfile
+import logging
+from typing import Any, Dict, Optional
+
+from langchain_core.callbacks import CallbackManagerForToolRun
+from langchain_core.tools import BaseTool
+from langchain_core.utils import get_from_dict_or_env
+from pydantic import model_validator
+
+logger = logging.getLogger(__name__)
+
+
+def _import_typecast() -> Any:
+    try:
+        import typecast
+    except ImportError as e:
+        raise ImportError(
+            "Cannot import typecast, please install `pip install typecast-python`."
+        ) from e
+    return typecast
+
+
+class TypecastText2SpeechTool(BaseTool):
+    """Tool that queries the Typecast Text2Speech API.
+
+    In order to set this up, follow instructions at:
+    https://typecast.ai/docs/overview
+    """
+
+    model: str = "ssfm-v21"
+    voice_id: str = "tc_62a8975e695ad26f7fb514d1"
+    language: Optional[str] = None
+    emotion_preset: str = "normal"
+    emotion_intensity: float = 1.0
+    audio_format: str = "wav"
+
+    name: str = "typecast_text2speech"
+    description: str = (
+        "A wrapper around Typecast Text2Speech. "
+        "Useful for when you need to convert text to speech. "
+        "It supports 27+ languages including English, Korean, Spanish, Japanese, "
+        "Chinese, and many more with emotion control capabilities. "
+    )
+
+    @model_validator(mode="before")
+    @classmethod
+    def validate_environment(cls, values: Dict) -> Any:
+        """Validate that api key exists in environment."""
+        _ = get_from_dict_or_env(values, "typecast_api_key", "TYPECAST_API_KEY")
+
+        return values
+
+    def _run(
+        self, query: str, run_manager: Optional[CallbackManagerForToolRun] = None
+    ) -> str:
+        """Use the tool."""
+        typecast = _import_typecast()
+
+        try:
+            # Get API key from environment
+            import os
+
+            api_key = os.environ.get("TYPECAST_API_KEY")
+
+            # Initialize client
+            client = typecast.Typecast(api_key=api_key)
+
+            # Create TTS request
+            request = typecast.TTSRequest(
+                text=query,
+                model=self.model,
+                voice_id=self.voice_id,
+                language=self.language,
+                prompt=typecast.Prompt(
+                    emotion_preset=self.emotion_preset,
+                    emotion_intensity=self.emotion_intensity,
+                ),
+                output=typecast.Output(audio_format=self.audio_format),
+            )
+
+            # Convert text to speech
+            response = client.text_to_speech(request)
+
+            # Save to temporary file
+            suffix = f".{self.audio_format}"
+            with tempfile.NamedTemporaryFile(
+                mode="wb", suffix=suffix, delete=False
+            ) as f:
+                f.write(response.audio_data)
+
+            return f.name
+        except Exception as e:
+            raise RuntimeError(f"Error while running TypecastText2SpeechTool: {e}")
+
+    def play(self, speech_file: str) -> None:
+        """Play the text as speech."""
+        try:
+            import sounddevice as sd
+            import soundfile as sf
+        except ImportError as e:
+            logger.error(
+                "Cannot import sounddevice or soundfile. "
+                "Please install them with `pip install sounddevice soundfile`. "
+                f"Error: {e}"
+            )
+            return
+
+        try:
+            data, samplerate = sf.read(speech_file)
+
+            # Get the current output device
+            output_device = sd.default.device[1]  # [input, output]
+
+            # Play on the current output device
+            sd.play(data, samplerate, device=output_device)
+            sd.wait()
+        except Exception as e:
+            logger.error(f"Error while playing audio: {e}")

libs/community/tests/integration_tests/tools/test_typecast.py

@@ -0,0 +1,172 @@
+"""Integration tests for Typecast Text2Speech Tool."""
+
+import os
+
+import pytest
+
+from langchain_community.tools.typecast import TypecastText2SpeechTool
+
+
+@pytest.mark.requires("typecast")
+class TestTypecastText2SpeechTool:
+    """Integration tests for TypecastText2SpeechTool."""
+
+    def test_typecast_text_to_speech_basic(self) -> None:
+        """Test basic text to speech conversion."""
+        # Skip if no API key is set
+        if not os.environ.get("TYPECAST_API_KEY"):
+            pytest.skip("TYPECAST_API_KEY not set")
+
+        tool = TypecastText2SpeechTool()
+        text = "Hello world! This is a test."
+
+        # Convert text to speech
+        audio_file = tool.run(text)
+
+        # Verify file was created
+        assert audio_file is not None
+        assert isinstance(audio_file, str)
+        assert os.path.exists(audio_file)
+        assert audio_file.endswith(".wav")
+
+        # Clean up
+        if os.path.exists(audio_file):
+            os.remove(audio_file)
+
+    def test_typecast_text_to_speech_with_emotion(self) -> None:
+        """Test text to speech with emotion settings."""
+        # Skip if no API key is set
+        if not os.environ.get("TYPECAST_API_KEY"):
+            pytest.skip("TYPECAST_API_KEY not set")
+
+        tool = TypecastText2SpeechTool(
+            emotion_preset="happy",
+            emotion_intensity=1.5,
+        )
+        text = "I am so excited to show you this!"
+
+        # Convert text to speech
+        audio_file = tool.run(text)
+
+        # Verify file was created
+        assert audio_file is not None
+        assert isinstance(audio_file, str)
+        assert os.path.exists(audio_file)
+
+        # Clean up
+        if os.path.exists(audio_file):
+            os.remove(audio_file)
+
+    def test_typecast_text_to_speech_mp3_format(self) -> None:
+        """Test text to speech with MP3 format."""
+        # Skip if no API key is set
+        if not os.environ.get("TYPECAST_API_KEY"):
+            pytest.skip("TYPECAST_API_KEY not set")
+
+        tool = TypecastText2SpeechTool(audio_format="mp3")
+        text = "This should be in MP3 format."
+
+        # Convert text to speech
+        audio_file = tool.run(text)
+
+        # Verify file was created
+        assert audio_file is not None
+        assert isinstance(audio_file, str)
+        assert os.path.exists(audio_file)
+        assert audio_file.endswith(".mp3")
+
+        # Clean up
+        if os.path.exists(audio_file):
+            os.remove(audio_file)
+
+    def test_typecast_text_to_speech_korean(self) -> None:
+        """Test text to speech with Korean language."""
+        # Skip if no API key is set
+        if not os.environ.get("TYPECAST_API_KEY"):
+            pytest.skip("TYPECAST_API_KEY not set")
+
+        tool = TypecastText2SpeechTool(language="kor")
+        text = "안녕하세요. 타입캐스트 테스트입니다."
+
+        # Convert text to speech
+        audio_file = tool.run(text)
+
+        # Verify file was created
+        assert audio_file is not None
+        assert isinstance(audio_file, str)
+        assert os.path.exists(audio_file)
+
+        # Clean up
+        if os.path.exists(audio_file):
+            os.remove(audio_file)
+
+    def test_typecast_with_agent(self) -> None:
+        """Test TypecastText2SpeechTool with a LangChain agent."""
+        # Skip if no API key is set
+        if not os.environ.get("TYPECAST_API_KEY") or not os.environ.get(
+            "OPENAI_API_KEY"
+        ):
+            pytest.skip("TYPECAST_API_KEY or OPENAI_API_KEY not set")
+
+        try:
+            from langchain.agents import create_agent
+            from langchain_openai import ChatOpenAI
+        except ImportError as e:
+            pytest.skip(f"Required packages not installed: {e}")
+
+        # Create LLM and tools
+        llm = ChatOpenAI(model="gpt-3.5-turbo-0125", temperature=0)
+        tools = [TypecastText2SpeechTool()]
+
+        # Create agent
+        agent = create_agent(
+            model=llm,
+            tools=tools,
+            system_prompt="You are a helpful assistant",
+        )
+
+        # Run the agent
+        query = (
+            "Convert this text to speech: 'Hello from LangChain agent!' "
+            "and tell me the file path."
+        )
+
+        # Invoke the agent
+        # The input schema for create_agent graph expects 'messages'
+        response = agent.invoke({"messages": [{"role": "user", "content": query}]})
+
+        # Verify response
+        # The output state contains 'messages'. The last message should be AIMessage.
+        messages = response["messages"]
+        last_message = messages[-1]
+        output = last_message.content
+        print(f"\nAgent response: {output}")
+
+        # The output should contain the file path or mention it
+        assert ".wav" in output or ".mp3" in output
+
+        # Try to clean up any audio files created during the test
+        import tempfile
+        import time
+
+        temp_dir = tempfile.gettempdir()
+        current_time = time.time()
+
+        # Find the file path in the output if possible
+        import re
+        match = re.search(r'/tmp/[\w-]+\.(wav|mp3)', output)
+        if match:
+            file_path = match.group(0)
+            if os.path.exists(file_path):
+                os.remove(file_path)
+        else:
+            # Fallback cleanup
+            for file in os.listdir(temp_dir):
+                if file.endswith((".wav", ".mp3")):
+                    file_path = os.path.join(temp_dir, file)
+                    try:
+                        # Only remove files created in the last 10 seconds
+                        if os.path.getmtime(file_path) > (current_time - 10):
+                            os.remove(file_path)
+                    except Exception:
+                        pass

libs/community/tests/unit_tests/tools/test_imports.py

@@ -133,6 +133,7 @@
     "TavilyAnswer",
     "TavilySearchResults",
     "Tool",
+    "TypecastText2SpeechTool",
     "VectorStoreQATool",
     "VectorStoreQAWithSourcesTool",
     "WikipediaQueryRun",