File size: 928 Bytes
a225ae4 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 |
import os
import whisper
from smolagents import Tool
class OpenAISpeechToTextTool(Tool):
"""
Tool to convert speech to text using OpenAI's Whisper model.
Args:
audio_path (str): Path to the audio file.
Returns:
str: Transcribed text from the audio file.
"""
name = "transcribe_audio"
description = "Transcribes audio to text and returns the text"
inputs = {
"audio_path": {"type": "string", "description": "Path to the audio file"},
}
output_type = "string"
def forward(self, audio_path: str) -> str:
try:
model = whisper.load_model("small")
if not os.path.exists(audio_path):
return f"Error: Audio file not found at {audio_path}"
result = model.transcribe(audio_path)
return result["text"]
except Exception as e:
return f"Error transcribing audio: {str(e)}"
|