MCP Server Whisper

"""MCP tools for audio processing operations.""" from typing import Optional from ..config import check_and_get_audio_path from ..constants import DEFAULT_MAX_FILE_SIZE_MB, SupportedChatWithAudioFormat from ..infrastructure import FileSystemRepository, MCPServer, SecurePathResolver from ..models import AudioProcessingResult from ..services import AudioService def create_audio_tools(mcp: MCPServer) -> None: """Register audio processing tools with the MCP server. Args: ---- mcp: FastMCP server instance. """ # Initialize services audio_path = check_and_get_audio_path() file_repo = FileSystemRepository(audio_path) path_resolver = SecurePathResolver(audio_path) audio_service = AudioService(file_repo, path_resolver) @mcp.tool(description="A tool used to convert audio files to mp3 or wav which are gpt-4o compatible.") async def convert_audio( input_file_name: str, target_format: SupportedChatWithAudioFormat = "mp3", output_file_name: Optional[str] = None, ) -> AudioProcessingResult: """Convert audio file to supported format (mp3 or wav). Args: input_file_name: Name of the input audio file to process target_format: Target audio format to convert to (mp3 or wav) output_file_name: Optional custom name for the output file. If not provided, defaults to input filename with appropriate extension Returns: ------- AudioProcessingResult with name of the converted audio file """ try: return await audio_service.convert_audio( input_filename=input_file_name, output_filename=output_file_name, target_format=target_format, ) except Exception as e: raise RuntimeError(f"Audio conversion failed for {input_file_name}: {str(e)}") from e @mcp.tool( description="A tool used to compress audio files which are >25mb. " "ONLY USE THIS IF THE USER REQUESTS COMPRESSION OR IF OTHER TOOLS FAIL DUE TO FILES BEING TOO LARGE." ) async def compress_audio( input_file_name: str, max_mb: int = DEFAULT_MAX_FILE_SIZE_MB, output_file_name: Optional[str] = None, ) -> AudioProcessingResult: """Compress audio file if it's larger than max_mb. Args: input_file_name: Name of the input audio file to process max_mb: Maximum file size in MB. Files larger than this will be compressed output_file_name: Optional custom name for the output file. If not provided, defaults to input filename with appropriate extension Returns: ------- AudioProcessingResult with name of the compressed audio file (or original if no compression needed) """ try: return await audio_service.compress_audio( input_filename=input_file_name, output_filename=output_file_name, max_mb=max_mb, ) except Exception as e: raise RuntimeError(f"Audio compression failed for {input_file_name}: {str(e)}") from e