agent_tools/AudioInterpreterTool.py

import os
from llama_index.readers.assemblyai import AssemblyAIAudioTranscriptReader
from llama_index.core.tools import FunctionTool

def transcribe_audio(audio_file_path: str) -> str:
    """
    Transcribes audio file and returns the transcript string as output.
    Args: 
    audio_file_path (str): Local path to the audio file
    """
    reader = AssemblyAIAudioTranscriptReader(api_key= os.getenv("ASSEMBLYAI_API_KEY") ,file_path=audio_file_path)
    docs = reader.load_data()
    return docs[0].text

def get_transcribe_audio() -> FunctionTool:
    return FunctionTool.from_defaults(
        fn=transcribe_audio,
        name="audio_file_transcriber",
        description="A tool that transcribes audio files (such as mp3). It takes the file_path to the downloaded audio file and returns the transcripts."
    )