Create video_translation.py

video_translation.py

@@ -0,0 +1,17 @@
+from smolagents.tools import Tool
+from huggingface_hub import InferenceClient
+
+class AudioTranscriptionTool(Tool):
+    name = "audio_transcription"
+    description = "Transcribes speech from an audio file (e.g., MP3)"
+    inputs = {"audio_path": {"type": "string", "description": "Path to the audio file"}}
+    output_type = "string"
+
+    def __init__(self):
+        super().__init__()
+        self.client = InferenceClient("openai/whisper-large-v3")
+
+    def forward(self, audio_path: str) -> str:
+        with open(audio_path, "rb") as f:
+            transcript = self.client.audio_to_text(f)
+        return transcript["text"]