Update app.py

app.py

@@ -9,12 +9,20 @@ import numpy as np
 
 
 
+
+
+
 def classify_audio(audio_file):
     model = AutoModelForAudioClassification.from_pretrained("3loi/SER-Odyssey-Baseline-WavLM-Multi-Attributes", trust_remote_code=True)
-    print(audio_file)
-    mean, std = -8.278621631819787e-05, 0.08485510250851999
-    raw_wav, _ = librosa.load(audio_file, sr=16000)
-    norm_wav = (raw_wav - mean) / (std+0.000001)
+    sr, raw_wav = audio_file
+    
+    print(audio_file, audio_file[1].dtype)
+    y = raw_wav.astype(np.float32)
+    y /= np.max(np.abs(y))
+
+    
+    #raw_wav, _ librosa.load(audio_file, sr=16000)
+    norm_wav = (y - mean) / (std+0.000001)
 
     mask = torch.ones(1, len(norm_wav))
     wavs = torch.tensor(norm_wav).unsqueeze(0)
@@ -25,11 +33,9 @@ def classify_audio(audio_file):
 
 
 def main():
-    audio_input = gr.inputs.Audio(source="upload", type="filepath")
-    output_text = gr.outputs.Textbox()
     
-    iface = gr.Interface(fn=classify_audio, inputs=audio_input, 
-                         outputs=output_text, title="Speech Emotion Recognition App",
+    iface = gr.Interface(fn=classify_audio, inputs=gr.Audio(sources=["upload", "microphone"], label="Audio file"), 
+                         outputs=gr.Text(), title="Speech Emotion Recognition App",
                          description="Upload an audio file and hit the 'Submit'\
                              button")
     
@@ -38,4 +44,4 @@ def main():
 
 if __name__ == '__main__':
     main()
-    
+