Update app.py

app.py

@@ -2,43 +2,41 @@ import gradio as gr
 import torch
 import numpy as np
 from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
+from scipy.signal import resample
 
-# Load model
+# Load model and processor
 model_id = "facebook/wav2vec2-large-960h-lv60-self"
 processor = Wav2Vec2Processor.from_pretrained(model_id)
 model = Wav2Vec2ForCTC.from_pretrained(model_id)
 
-# Transcription function
-def transcribe(audio_np, sample_rate):
-    if audio_np is None:
-        return "No audio received."
+# Transcribe function
+def transcribe(audio, sample_rate):
+    if audio is None:
+        return "⚠️ No audio received."
 
-    # Resample to 16kHz if needed
+    # Resample audio to 16kHz if needed
     if sample_rate != 16000:
-        import torchaudio
-        audio_tensor = torch.tensor(audio_np).unsqueeze(0)
-        resampler = torchaudio.transforms.Resample(orig_freq=sample_rate, new_freq=16000)
-        audio_tensor = resampler(audio_tensor)
-    else:
-        audio_tensor = torch.tensor(audio_np).unsqueeze(0)
+        number_of_samples = round(len(audio) * float(16000) / sample_rate)
+        audio = resample(audio, number_of_samples)
 
-    input_values = processor(audio_tensor.squeeze(), sampling_rate=16000, return_tensors="pt").input_values
+    # Prepare input
+    input_values = processor(audio, sampling_rate=16000, return_tensors="pt").input_values
 
+    # Run model
     with torch.no_grad():
         logits = model(input_values).logits
+        predicted_ids = torch.argmax(logits, dim=-1)
+        transcription = processor.batch_decode(predicted_ids)[0]
 
-    predicted_ids = torch.argmax(logits, dim=-1)
-    transcription = processor.batch_decode(predicted_ids)[0]
     return transcription.lower()
 
-# Interface
+# Gradio interface
 demo = gr.Interface(
     fn=transcribe,
-    inputs=gr.Audio(source="microphone", type="numpy", label="Speak now"),
-    outputs=gr.Textbox(label="Transcription"),
-    live=False,
+    inputs=gr.Audio(source="microphone", type="numpy", label="🎤 Speak now"),
+    outputs=gr.Textbox(label="📝 Transcription"),
     title="Wav2Vec2 Speech Transcription",
-    description="Speak into your mic and get a transcription using Wav2Vec2!"
+    description="Speak and get real-time transcription using Wav2Vec2 (Hugging Face)."
 )
 
 demo.launch()