Update app.py

app.py

@@ -1,5 +1,8 @@
 import streamlit as st
 import speech_recognition as sr
+import os
+from pydub import AudioSegment
+from tempfile import NamedTemporaryFile
 
 def transcribe_audio(audio_file):
     recognizer = sr.Recognizer()
@@ -15,18 +18,34 @@ def transcribe_audio(audio_file):
 
 def filter_audio(audio_file, input_text):
     transcribed_text = transcribe_audio(audio_file)
-    transcribed_words = set(transcribed_text.split())
-    input_words = set(input_text.split())
+    transcribed_words = transcribed_text.split()
+    input_words = input_text.split()
 
-    # Retain only the words in the transcribed text that match the input text
-    filtered_words = transcribed_words.intersection(input_words)
+    filtered_words = [word for word in transcribed_words if word in input_words]
 
-    # Create a new transcription with the filtered words
-    filtered_transcription = " ".join(filtered_words)
+    return " ".join(filtered_words)
 
-    return filtered_transcription
+def create_filtered_audio(original_audio, matching_words):
+    original_audio = AudioSegment.from_file(original_audio, format="wav")
 
-st.title("Audio Word Filter")
+    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")
@@ -36,10 +55,13 @@ if uploaded_audio is not None:
 
 if st.button("Filter Audio"):
     if uploaded_audio is not None:
-        filtered_text = filter_audio(uploaded_audio, input_text)
-        st.subheader("Filtered Audio Transcript:")
-        st.write(filtered_text)
+        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 and filters it to retain only the words mentioned in the input text."
+    "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."
 )