Update app.py

app.py

@@ -1,40 +1,21 @@
-import gradio as gr
-from transformers import Wav2Vec2ForSequenceClassification, Wav2Vec2Processor
-import torch
-import torchaudio
+model_name = "Dpngtm/wav2vec2-emotion-recognition"
+processor  = Wav2Vec2Processor.from_pretrained(model_name)
+model      = Wav2Vec2ForSequenceClassification.from_pretrained(model_name)
 
-# Load the pre-trained model and processor
-model_name = "bhadresh-savani/wav2vec2-large-robust-english-emotion"
-processor = Wav2Vec2Processor.from_pretrained(model_name)
-model = Wav2Vec2ForSequenceClassification.from_pretrained(model_name)
-
-# Emotion labels for this specific model
-labels = ['angry', 'calm', 'happy', 'sad']
+labels = ["angry", "calm", "disgust", "fearful", "happy", "neutral", "sad", "surprised"]
 
 def predict_emotion(audio):
-    # audio: tuple -> (sample_rate, numpy array)
     speech, sr = audio
     if sr != 16000:
         resampler = torchaudio.transforms.Resample(sr, 16000)
-        speech = resampler(torch.tensor(speech))
+        speech    = resampler(torch.tensor(speech))
     else:
-        speech = torch.tensor(speech)
+        speech    = torch.tensor(speech)
 
     input_values = processor(speech, sampling_rate=16000, return_tensors="pt").input_values
     with torch.no_grad():
         logits = model(input_values).logits
 
     predicted_id = torch.argmax(logits, dim=-1).item()
-    emotion = labels[predicted_id]
+    emotion      = labels[predicted_id]
     return f"Predicted Emotion: **{emotion}**"
-
-# Gradio interface
-interface = gr.Interface(
-    fn=predict_emotion,
-    inputs=gr.Audio(source="microphone", type="numpy", label="Record or Upload Speech"),
-    outputs=gr.Markdown(label="Emotion"),
-    title="Voice Emotion Recognition",
-    description="Speak or upload a WAV file to detect the emotion using a fine-tuned Wav2Vec2 model."
-)
-
-interface.launch()