Update app.py

app.py

@@ -1,22 +1,14 @@
 # Importing the required libraries
-import os
-os.system('apt-get update && apt-get install -y portaudio19-dev')
-
 import streamlit as st
-import sounddevice as sd
 import numpy as np
-from scipy.io.wavfile import write
 import librosa
 import matplotlib.pyplot as plt
 from tensorflow.keras.models import Sequential, load_model
 from tensorflow.keras.layers import Dense, Dropout
-import io
-
-print(sd.query_devices())
-
-# Set the device ID (replace 'device_id' with the actual ID)
-device_id = 1  # Example: choose the correct device ID from the list
-sd.default.device = device_id
+import io 
+import soundfile as sf
+from streamlit_audio_recorder import st_audiorec
+from scipy.io.wavfile import write, read as wav_read
 
 # Define the target speakers in a list.
 # ALSO THESE ARE FILES THAT WERE NOT USED TO TEST NOR TRAIN THE MODEL
@@ -150,49 +142,45 @@ if uploaded_file is not None:
             speaker, probability, _ = classify_speaker("uploaded_audio.wav")
             gender, gen_probability = classify_gender("uploaded_audio.wav")
             
-            # Whats the gender of speaker?
+            # What's the gender of speaker?
             st.write(f"Predicted Gender: {gender}")
 
-            # Whats the shot of speaker being a male or female
+            # What's the shot of speaker being a male or female
             st.write(f"Gender Probability: {gen_probability}")
 
             # Which speaker is it?
             st.write(f"Predicted Speaker: {speaker}")
 
-            # Whats the chances of being spealer?
+            # What's the chances of being the speaker?
             st.write(f"Speaker Probability: {probability}")
             
         except Exception as e:
             st.error(f"Error occurred: {e}")
 
-# Record audio with sounddevice
-def record_audio(duration=8, samplerate=22050):
-    st.write("Recording...")
-    audio_data = sd.rec(int(duration * samplerate), samplerate=samplerate, channels=1, dtype='int16')
-    sd.wait()  # Wait until recording is finished
-    write("recorded_audio.wav", samplerate, audio_data)  # Save as WAV file
-    return "recorded_audio.wav"
-
-if st.button("Record Audio"):
-    wav_file_path = record_audio(duration=8)
-    st.write(f"Audio recorded and saved to {wav_file_path}") # save auido
-    st.audio(wav_file_path) # show audio
-    audio_to_spectrogram(wav_file_path) # melspectogram of recorded audio
+# Record audio with streamlit_audio_recorder
+recorded_audio = st_audiorec()
+
+if recorded_audio:
+    # Save the audio as a .wav file
+    with open("recorded_audio.wav", "wb") as f:
+        f.write(recorded_audio)
+
+    st.write(f"Audio recorded and saved to recorded_audio.wav")  # Show message
+    st.audio("recorded_audio.wav")  # Show the audio file
+
+    # Process the recorded audio
+    audio_to_spectrogram("recorded_audio.wav")
     st.image("spectrogram.png", caption="Mel Spectrogram of the recorded audio file", use_column_width="auto", width=200)
-    speaker, probability, wav_file = classify_speaker(wav_file_path)
-    gender, gen_probability = classify_gender(wav_file_path)
+
+    # Classify the speaker and gender
+    speaker, probability, wav_file = classify_speaker("recorded_audio.wav")
+    gender, gen_probability = classify_gender("recorded_audio.wav")
     
-    # who do u sound like the most?
+    # Display results
     st.write(f"Predicted Gender: {gender}")
-    
-    # how much do u sound like this speaker?
     st.write(f"Gender Probability: {gen_probability}")
-    
-    # who do u sound like the most?
     st.write(f"Predicted Speaker: {speaker}")
-    
-    # how much do u sound like this speaker?
     st.write(f"Speaker Probability: {probability}")
     
-    # speaker audio
+    # Display the wav file of the predicted speaker
     st.audio(wav_file)