Update app.py

app.py

@@ -13,14 +13,20 @@ def transcribe_audio(audio_file):
     except sr.UnknownValueError:
         return ""
 
-def filter_words(input_text, transcribed_text):
-    transcribed_words = transcribed_text.split()
-    filtered_text = input_text
-    for word in transcribed_words:
-        filtered_text = filtered_text.replace(word, "")
-    return filtered_text
+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())
 
-st.title("Voice Cloning Word Filter")
+    # Retain only the words in the transcribed text that match the input text
+    filtered_words = transcribed_words.intersection(input_words)
+
+    # Create a new transcription with the filtered words
+    filtered_transcription = " ".join(filtered_words)
+
+    return filtered_transcription
+
+st.title("Audio Word Filter")
 
 uploaded_audio = st.file_uploader("Upload an audio file", type=["wav", "mp3", "ogg"])
 input_text = st.text_area("Enter input text")
@@ -28,16 +34,12 @@ input_text = st.text_area("Enter input text")
 if uploaded_audio is not None:
     st.audio(uploaded_audio, format="audio/wav")
 
-if st.button("Transcribe and Filter"):
+if st.button("Filter Audio"):
     if uploaded_audio is not None:
-        transcribed_text = transcribe_audio(uploaded_audio)
-        filtered_text = filter_words(input_text, transcribed_text)
-        st.subheader("Transcribed Text:")
-        st.write(transcribed_text)
-        st.subheader("Filtered Output Text:")
+        filtered_text = filter_audio(uploaded_audio, input_text)
+        st.subheader("Filtered Audio Transcript:")
         st.write(filtered_text)
 
 st.write(
-    "Note: This is a simple demonstration of filtering words from an audio transcript. The accuracy of word filtering "
-    "depends on the quality of the audio and the performance of the ASR engine."
+    "Note: This app transcribes the uploaded audio and filters it to retain only the words mentioned in the input text."
 )