Update app.py

app.py

@@ -1,69 +1,35 @@
 import streamlit as st
-import speech_recognition as sr
-import os
+import noisereduce as nr
 from pydub import AudioSegment
-from tempfile import NamedTemporaryFile
 
-def transcribe_audio(audio_file):
-    recognizer = sr.Recognizer()
+# Define a Streamlit app
+st.title("Audio Processing App")
 
-    with sr.AudioFile(audio_file) as source:
-        audio = recognizer.record(source)
-
-    try:
-        transcription = recognizer.recognize_google(audio)
-        return transcription
-    except sr.UnknownValueError:
-        return ""
-
-def filter_audio(audio_file, input_text):
-    transcribed_text = transcribe_audio(audio_file)
-    transcribed_words = transcribed_text.split()
-    input_words = input_text.split()
-
-    matching_words = [word for word in transcribed_words if word in input_words]
-
-    return " ".join(matching_words)
-
-def create_filtered_audio(original_audio, matching_words):
-    transcribed_text = filter_audio(original_audio, matching_words)
-
-    original_audio = AudioSegment.from_file(original_audio, format="wav")
-
-    matching_segments = []
-    matching_words = matching_words.split()
-
-    for word in matching_words:
-        word_start = transcribed_text.index(word)
-        word_end = word_start + len(word)
-        matching_segments.append(original_audio[word_start * 1000 : word_end * 1000])
-
-    filtered_audio = AudioSegment.silent(duration=len(original_audio))
-    for segment in matching_segments:
-        filtered_audio = filtered_audio.overlay(segment)
-
-    with NamedTemporaryFile(suffix=".wav", delete=False) as temp_audio_file:
-        filtered_audio.export(temp_audio_file.name, format="wav")
-
-    return temp_audio_file.name
-
-st.title("Audio Word Filter and Regeneration")
-
-uploaded_audio = st.file_uploader("Upload an audio file", type=["wav", "mp3", "ogg"])
-input_text = st.text_area("Enter input text")
+# Upload the input audio file
+uploaded_audio = st.file_uploader("Upload an audio file", type=["mp3", "wav"])
 
 if uploaded_audio is not None:
-    st.audio(uploaded_audio, format="audio/wav")
-
-if st.button("Filter Audio"):
-    if uploaded_audio is not None:
-        matching_words = filter_audio(uploaded_audio, input_text)
-        st.subheader("Matching Words:")
-        st.write(matching_words)
-        filtered_audio = create_filtered_audio(uploaded_audio, matching_words)
-        st.subheader("Filtered Audio:")
-        st.audio(filtered_audio, format="audio/wav")
-
-st.write(
-    "Note: This app transcribes the uploaded audio, filters it to retain only the words mentioned in the input text, and generates filtered audio without altering the original audio."
-)
+    # Clean the input audio using noisereduce
+    # You may need to adjust the parameters for your specific audio
+    st.write("Cleaning audio...")
+    # Load the audio file
+    audio = AudioSegment.from_file(uploaded_audio)
+    # Convert to numpy array for processing
+    audio_data = audio.get_array_of_samples()
+    # Apply noise reduction
+    reduced_audio = nr.reduce_noise(y=audio_data, sr=audio.frame_rate)
+
+    # Slow down the audio
+    st.write("Slowing down audio...")
+    slowed_audio = audio.speedup(playback_speed=0.7)  # Adjust the speed factor as needed
+
+    # Generate a new audio file
+    st.write("Generating output audio...")
+    slowed_audio.export("output_audio.wav", format="wav")
+
+    # Provide the download link for the cleaned and slowed audio
+    st.audio("output_audio.wav")
+
+# Run the Streamlit app
+if __name__ == "__main__":
+    st.write("Upload an audio file to get started.")