Update app.py

app.py

@@ -9,6 +9,7 @@ from profanity_check import predict, predict_prob
 from pydub import AudioSegment
 from pydub.playback import play
 from subprocess import Popen, PIPE
+import gradio as gr
 
 def parse_args(): 
     """ 
@@ -212,72 +213,21 @@ def mask_profanities(vocal_stem, profanities):
     return vocals
 
 
+def process_audio(input_file, model_size):
+    args = parse_args()
+    inpath = os.path.abspath(input_file.name)
+    outpath, vocal_stem, instr_stem = main(args, input_file=inpath, model_size=model_size)
+    return outpath
+
 if __name__ == "__main__":
     args = parse_args()
     
     if len(sys.argv)>1:
         main(args, skip_ss=False)
     else:
-        import streamlit as st
-        st.title('Saylss')
-        with st.expander("About", expanded=False):
-            st.markdown('''
-            This app processes an input audio track (.mp3 or .wav) with the purpose of identifying and muting profanities in the song.
-            
-
-            A larger model takes longer to run and is more accurate, and vice-versa.
-
-
-            Simply select the model size and upload your file!
-            ''')
-        model = st.selectbox('Choose model size:', ('tiny','small','medium'), index=1)
-
-        uploaded_file = st.file_uploader(
-            "Choose input track:",
-            type=[".mp3",".wav"],
-            accept_multiple_files=False,
+        iface = gr.Interface(
+            fn=process_audio, 
+            inputs=[gr.inputs.Audio(source="upload"), gr.inputs.Radio(['tiny','small','medium'])], 
+            outputs=gr.outputs.Audio(type="file")
         )
-
-        if uploaded_file is not None:
-            uploaded_file.name = uploaded_file.name.replace(' ','_')
-            ext = os.path.splitext(uploaded_file.name)[1]
-            if ext == '.wav':
-                st_format = 'audio/wav'
-            elif ext == '.mp3':
-                st_format = 'audio/mp3'
-
-            uploaded_file_content = uploaded_file.getvalue()
-            with open(uploaded_file.name, 'wb') as f:
-                f.write(uploaded_file_content)
-
-            audio_bytes_input = uploaded_file_content
-            st.audio(audio_bytes_input, format=st_format)
-            
-            # run code
-            with st.spinner('Processing input audio...'):
-                inpath = os.path.abspath(uploaded_file.name)
-                outpath, vocal_stem, instr_stem = main(args, input_file=inpath, model_size=model)
-                
-                if outpath == 'No profanities found':
-                    st.text(outpath + ' - Refresh the page and try a different song or model size')
-                    sys.exit()
-            
-            # display output audio
-            #st.text('Play output Track:')
-            st.text('\nOutput:')
-            audio_file = open(outpath, 'rb')
-            audio_bytes = audio_file.read()
-            st.audio(audio_bytes, format=st_format)
-
-            # flush all media
-            if os.path.isfile(inpath):
-                os.remove(inpath)
-            if os.path.isfile(outpath):
-                os.remove(outpath)
-            if os.path.isfile(vocal_stem):
-                os.remove(vocal_stem)
-            if os.path.isfile(instr_stem):
-                os.remove(instr_stem)
-            sep_dir = os.path.split(instr_stem)[0]
-            if os.path.isdir(sep_dir):
-                os.rmdir(sep_dir)
+        iface.launch()