All Voice Lab MCP Server

Instructions

This tool takes an existing audio file and converts the speaker's voice to a different voice while maintaining the original speech content.

Args:
    audio_file_path: Path to the source audio file. Only MP3 and WAV formats are supported. Maximum file size: 50MB.
    voice_id: Voice ID to use for the conversion. Required. Must be a valid voice ID from the available voices (use get_voices tool to retrieve).
    similarity: Voice similarity factor, range [0, 1], where 0 is least similar and 1 is most similar to the original voice characteristics. Default value is 1.
    remove_background_noise: Whether to remove background noise from the source audio before conversion. Default is False.
    output_dir: Output directory for the generated audio file. Default is user's desktop.
    
Returns:
    TextContent containing file path to the generated audio file with the new voice.
    
Limitations:
    - Only MP3 and WAV formats are supported
    - Maximum file size: 50MB
    - File must exist and be accessible

Implementation Reference

• allvoicelab_mcp/tools/speech.py:105-168 (handler)

  The main handler function for the speech_to_speech tool. It validates inputs, calls the AllVoiceLab client to perform the voice conversion, and returns the output file path or error.

  def speech_to_speech(
      audio_file_path: str,
      voice_id: str,
      similarity: float = 1,
      remove_background_noise: bool = False,
      output_dir: str = None
  ) -> TextContent:
      """
      Convert audio to another voice while preserving speech content
      
      Args:
          audio_file_path: Path to the source audio file. Only MP3 and WAV formats are supported. Maximum file size: 50MB.
          voice_id: Voice ID to use for the conversion. Required. Must be a valid voice ID from the available voices (use get_voices tool to retrieve).
          similarity: Voice similarity factor, range [0, 1], where 0 is least similar and 1 is most similar to the original voice characteristics. Default value is 1.
          remove_background_noise: Whether to remove background noise from the source audio before conversion. Default is False.
          output_dir: Output directory for the generated audio file. Default is user's desktop.
          
      Returns:
          TextContent: Contains the file path to the generated audio file with the new voice.
      """
      all_voice_lab = get_client()
      output_dir = all_voice_lab.get_output_path(output_dir)
      logging.info(f"Tool called: speech_to_speech, voice_id: {voice_id}, similarity: {similarity}")
      logging.info(f"Audio file path: {audio_file_path}, remove background noise: {remove_background_noise}")
      logging.info(f"Output directory: {output_dir}")
  
      # Validate audio file
      is_valid, error_message = validate_audio_file(audio_file_path)
      if not is_valid:
          return create_error_response(error_message)
  
      # Validate voice_id parameter
      if not voice_id:
          logging.warning("voice_id parameter is empty")
          return create_error_response("voice_id parameter cannot be empty")
  
      # Validate voice_id format (basic check)
      if not isinstance(voice_id, str) or len(voice_id.strip()) == 0:
          logging.warning(f"Invalid voice_id format: {voice_id}")
          return create_error_response("Invalid voice_id format")
  
      # Validate similarity range
      if similarity < 0 or similarity > 1:
          logging.warning(f"Similarity parameter {similarity} is out of range [0, 1]")
          return create_error_response("similarity parameter must be between 0 and 1")
  
      # Validate and create output directory
      is_valid, error_message = validate_output_directory(output_dir)
      if not is_valid:
          return create_error_response(error_message)
  
      try:
          logging.info("Starting speech conversion processing")
          file_path = all_voice_lab.speech_to_speech(audio_file_path, voice_id, output_dir, similarity,
                                                     remove_background_noise)
          logging.info(f"Speech conversion successful, file saved at: {file_path}")
          return create_success_response(f"Audio conversion completed, file saved at: {file_path}\n")
      except FileNotFoundError as e:
          logging.error(f"Audio file does not exist: {audio_file_path}, error: {str(e)}")
          return create_error_response(f"Audio file does not exist: {audio_file_path}")
      except Exception as e:
          logging.error(f"Speech conversion failed: {str(e)}")
          return create_error_response("Conversion failed, tool temporarily unavailable")

• allvoicelab_mcp/server.py:78-99 (registration)

  The MCP tool registration for speech_to_speech, including name, description with input schema, and binding to the handler function.

  mcp.tool(
      name="speech_to_speech",
      description="""[AllVoiceLab Tool] Convert audio to another voice while preserving speech content.
      
      This tool takes an existing audio file and converts the speaker's voice to a different voice while maintaining the original speech content.
      
      Args:
          audio_file_path: Path to the source audio file. Only MP3 and WAV formats are supported. Maximum file size: 50MB.
          voice_id: Voice ID to use for the conversion. Required. Must be a valid voice ID from the available voices (use get_voices tool to retrieve).
          similarity: Voice similarity factor, range [0, 1], where 0 is least similar and 1 is most similar to the original voice characteristics. Default value is 1.
          remove_background_noise: Whether to remove background noise from the source audio before conversion. Default is False.
          output_dir: Output directory for the generated audio file. Default is user's desktop.
          
      Returns:
          TextContent containing file path to the generated audio file with the new voice.
          
      Limitations:
          - Only MP3 and WAV formats are supported
          - Maximum file size: 50MB
          - File must exist and be accessible
      """
  )(speech_to_speech)