Update app.py

app.py

@@ -3,70 +3,51 @@ import numpy as np
 import soundfile as sf
 import gradio as gr
 
-def calculate_pesq(audio_ref, audio_deg):
-    # Carica i file audio di riferimento e degradato
-    ref_signal, ref_sample_rate = sf.read(audio_ref)
-    deg_signal, deg_sample_rate = sf.read(audio_deg)
+def calculate_pesq(reference_audio, degraded_audio):
+    # Directly use the audio data and sample rates passed from Gradio
+    ref_signal, ref_sample_rate = reference_audio
+    deg_signal, deg_sample_rate = degraded_audio
 
-    # Assicurati che i file audio abbiano la stessa frequenza di campionamento
+    # Convert audio data to the same sample rate if needed
     if ref_sample_rate != deg_sample_rate:
-        print("I due file audio hanno frequenze di campionamento diverse. Eseguo il resampling.")
         from scipy.signal import resample
         min_length = min(len(ref_signal), len(deg_signal))
         ref_signal = resample(ref_signal[:min_length], len(ref_signal))
         deg_signal = resample(deg_signal[:min_length], len(deg_signal))
 
-    # Calcola il punteggio PESQ tra i due file audio
     pesq_score = pesq.pesq(ref_signal, deg_signal, ref_sample_rate)
-
     return pesq_score
 
-def calculate_snr(audio_ref, audio_deg):
-    # Carica i file audio di riferimento e degradato
-    ref_signal, _ = sf.read(audio_ref)
-    deg_signal, _ = sf.read(audio_deg)
+def calculate_snr(reference_audio, degraded_audio):
+    ref_signal, _ = reference_audio
+    deg_signal, _ = degraded_audio
 
-    # Calcola l'SNR tra i due file audio
     noise = ref_signal - deg_signal
     snr = 10 * np.log10(np.sum(ref_signal**2) / np.sum(noise**2))
-
     return snr
 
-def calculate_audio_difference(audio_ref, audio_deg):
-    # Carica i file audio di riferimento e degradato
-    ref_signal, ref_sample_rate = sf.read(audio_ref)
-    deg_signal, deg_sample_rate = sf.read(audio_deg)
+def calculate_audio_difference(reference_audio, degraded_audio):
+    ref_signal, ref_sample_rate = reference_audio
+    deg_signal, deg_sample_rate = degraded_audio
 
-    # Verifica e correggi le differenze nella frequenza di campionamento
     if ref_sample_rate != deg_sample_rate:
-        print("I due file audio hanno frequenze di campionamento diverse. Eseguo il resampling.")
         from scipy.signal import resample
         min_length = min(len(ref_signal), len(deg_signal))
         ref_signal = resample(ref_signal[:min_length], len(ref_signal))
         deg_signal = resample(deg_signal[:min_length], len(deg_signal))
 
-    # Verifica e correggi le differenze nella lunghezza
-    if len(ref_signal) != len(deg_signal):
-        print("I due file audio hanno lunghezze diverse. Eseguo il padding.")
-        max_length = max(len(ref_signal), len(deg_signal))
-        ref_signal = np.pad(ref_signal, (0, max_length - len(ref_signal)))
-        deg_signal = np.pad(deg_signal, (0, max_length - len(deg_signal)))
-
-    # Calcola la differenza tra i segnali audio campione per campione
     diff_count = np.count_nonzero(ref_signal != deg_signal)
-
-    # Calcola la percentuale di differenza tra i due segnali audio
     total_samples = len(ref_signal)
     difference_percentage = (diff_count / total_samples) * 100
 
     return difference_percentage
 
 def audio_quality_interface(reference_audio, degraded_audio):
-    pesq_score = calculate_pesq(reference_audio.name, degraded_audio.name)
-    snr = calculate_snr(reference_audio.name, degraded_audio.name)
-    difference_percentage = calculate_audio_difference(reference_audio.name, degraded_audio.name)
+    pesq_score = calculate_pesq(reference_audio, degraded_audio)
+    snr = calculate_snr(reference_audio, degraded_audio)
+    difference_percentage = calculate_audio_difference(reference_audio, degraded_audio)
 
-    result = f"Risultati:\n\nPunteggio PESQ: {pesq_score}\nSNR: {snr}\nPercentuale di differenza: {difference_percentage:.2f}%"
+    result = f"Risultati:\n\nPunteggio PESQ: {pesq_score:.2f}\nSNR: {snr:.2f}\nPercentuale di differenza: {difference_percentage:.2f}%"
     return result
 
 iface = gr.Interface(fn=audio_quality_interface,