Update app.py

app.py

@@ -1,8 +1,10 @@
 import streamlit as st
+import noisereduce as nr
 import soundfile as sf
 import io
 import numpy as np
 from pydub import AudioSegment
+from pydub.silence import split_on_silence, detect_silence
 
 # Define a Streamlit app
 st.title("Audio Processing App")
@@ -13,26 +15,30 @@ uploaded_audio = st.file_uploader("Upload an audio file", type=["mp3", "wav", "o
 if uploaded_audio is not None:
     audio_bytes = uploaded_audio.read()
 
-    # Convert audio file to numpy array using soundfile
+    # Convert audio file to numpy array
     audio, sample_rate = sf.read(io.BytesIO(audio_bytes))
 
-    # Convert the audio to 16-bit to avoid unsupported 24-bit formats
-    audio = (audio * 32767).astype(np.int16)
+    # Apply noise reduction
+    st.write("Applying noise reduction...")
+    reduced_audio_data = nr.reduce_noise(y=audio, sr=sample_rate)
 
-    # Create an AudioSegment from the audio data
-    audio_segment = AudioSegment(
-        audio.tobytes(),
+    # Create an AudioSegment from the reduced audio data
+    reduced_audio = AudioSegment(
+        reduced_audio_data.tobytes(),
         frame_rate=sample_rate,
-        sample_width=2,
+        sample_width=reduced_audio_data.dtype.itemsize,
         channels=1
     )
 
-    # Slow down the audio based on user's input speed factor
-    st.write("Slowing down audio...")
-    slowed_audio = audio_segment.speedup(playback_speed=0.7)
+    # Split audio on silences
+    st.write("Inserting small silences...")
+    silence_segments = detect_silence(reduced_audio, min_silence_len=100, silence_thresh=-36)
+    silenced_audio = AudioSegment.empty()
+    for start, end in silence_segments:
+        silenced_audio += reduced_audio[start:end]
 
     # Provide a link to download the processed audio
-    st.audio(slowed_audio.export(format="wav").read(), format="audio/wav")
+    st.audio(silenced_audio.export(format="wav").read(), format="audio/wav")
 
 # Run the Streamlit app
 if __name__ == "__main__":