Update app.py

app.py

@@ -1,23 +1,24 @@
 import gradio as gr
+import librosa
+import librosa.display
 import matplotlib.pyplot as plt
-import numpy as np
-import os
-import soundfile as sf
 
 def create_spectrogram_and_get_info(audio_file):
-
-    # Clear figure in case it has data in it
-    plt.clf()
-    
     # Read the audio data from the file
-    audio_data, sample_rate = sf.read(audio_file)
+    audio_data, sample_rate = librosa.load(audio_file)
+
+    # Compute the mel-scaled spectrogram
+    spectrogram = librosa.feature.melspectrogram(y=audio_data, sr=sample_rate)
 
-    # Flatten the audio data if it's not mono
-    audio_data = audio_data.flatten() if len(audio_data.shape) > 1 else audio_data
+    # Convert the power spectrogram to decibel (dB) units
+    spectrogram_db = librosa.power_to_db(spectrogram, ref=np.max)
 
-    # Create the spectrogram
-    plt.specgram(audio_data, Fs=sample_rate / 1, NFFT=4096, sides='onesided',
-                 cmap="Reds_r", scale_by_freq=True, scale='dB', mode='magnitude')
+    # Display the spectrogram
+    plt.figure(figsize=(10, 4))
+    librosa.display.specshow(spectrogram_db, x_axis='time', y_axis='mel', sr=sample_rate, fmax=8000)
+    plt.colorbar(format='%+2.0f dB')
+    plt.title('Mel spectrogram')
+    plt.tight_layout()
 
     # Save the spectrogram to a PNG file
     plt.savefig('spectrogram.png')
@@ -40,9 +41,3 @@ def create_spectrogram_and_get_info(audio_file):
 
     # Return the PNG file of the spectrogram and the info table
     return info_table, 'spectrogram.png'
-
-# Create the Gradio interface
-iface = gr.Interface(fn=create_spectrogram_and_get_info, inputs=gr.Audio(type="filepath"), outputs=["markdown", "image"])
-iface.launch()
-
-