Update tools.py

tools.py

@@ -146,7 +146,7 @@ class audio_or_mp3__interpreter(Tool):
     name="audio_tool"
     description = "Allows you to convert audio into text. It uses Whisper, it is a state-of-the-art model for automatic speech recognition (ASR) and speech translation"
     inputs = {
-        'audio': {"type": "audio", "description": "the audio of interest"}
+        'audio': {"type": "audio", "description": "the audio of interest. Must be in the format mp3."}
     }
     output_type = "string"
     
@@ -155,12 +155,8 @@ class audio_or_mp3__interpreter(Tool):
         torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
         
         model_id = "openai/whisper-large-v3"
-        
-        model = AutoModelForSpeechSeq2Seq.from_pretrained(
-            model_id, torch_dtype=torch_dtype, low_cpu_mem_usage=True, use_safetensors=True
-        )
+        model = AutoModelForSpeechSeq2Seq.from_pretrained(model_id, torch_dtype=torch_dtype)
         model.to(device)
-        
         processor = AutoProcessor.from_pretrained(model_id)
         
         pipe = pipeline(
@@ -169,14 +165,14 @@ class audio_or_mp3__interpreter(Tool):
             tokenizer=processor.tokenizer,
             feature_extractor=processor.feature_extractor,
             torch_dtype=torch_dtype,
-            device=device,
+            device=device
         )
-        mp3_file = AudioSegment.from_mp3({audio})
-        wav_file = mp3_file.export("output.wav", format="wav")
-        sample_rate, audio_data = wavfile.read("output.wav")
-        audio_data = audio_data / 32768.0
         
-        result = pipe(audio_data)
+        audio = AudioSegment.from_mp3({audio})
+        audio = audio.set_channels(1).set_frame_rate(16000)
+        audio.export("output.wav", format="wav")
+        
+        result = pipe("output.wav", return_timestamps=True)
         return result["text"]
 
 class Wikipedia_reader(Tool):