Update app.py

app.py

@@ -1,6 +1,6 @@
 from transformers import AutoFeatureExtractor
 from transformers import AutoModelForAudioClassification
-import torchaudio
+import librosa
 from detect_face import detect_face
 from transformers import AutoModelForImageClassification
 from transformers import AutoImageProcessor
@@ -71,18 +71,14 @@ def predict_with_model(image, model, processor):
         "confidence": round(confidence * 100, 2)
     }
 
-def predict_audio(audio_path):
 
-    torchaudio.set_audio_backend("soundfile")
+def predict_audio(audio_path):
 
-    waveform, sample_rate = torchaudio.load(
-    audio_path,
-    backend="soundfile"
-    )
+    waveform, sr = librosa.load(audio_path, sr=16000)
 
     inputs = voice_processor(
-        waveform.squeeze().numpy(),
-        sampling_rate=sample_rate,
+        waveform,
+        sampling_rate=16000,
         return_tensors="pt"
     )
 
@@ -90,13 +86,9 @@ def predict_audio(audio_path):
         outputs = voice_model(**inputs)
 
     logits = outputs.logits
-
     predicted_class = logits.argmax(-1).item()
 
-    confidence = torch.softmax(
-        logits,
-        dim=1
-    )[0][predicted_class].item()
+    confidence = torch.softmax(logits, dim=1)[0][predicted_class].item()
 
     label = voice_model.config.id2label[predicted_class]