Update app.py

app.py

@@ -1,12 +1,10 @@
+from warnings import filterwarnings
 import streamlit as st
-import whisper
-from sklearn.feature_extraction.text import TfidfVectorizer
-from sklearn.metrics.pairwise import cosine_similarity
 import base64
 from pydub import AudioSegment
-from hezar.models import Model
-import librosa
-import soundfile as sf
+from functions import *
+filterwarnings('ignore')
+
 
 st.set_page_config(
     page_title="Sing It Forward App",
@@ -68,77 +66,7 @@ st.write('------')
 
 
 
-def cosine_sim(text1, text2):
-    vectorizer = TfidfVectorizer().fit_transform([text1, text2])
-    vectors = vectorizer.toarray()
-    return cosine_similarity(vectors)[0, 1]
-    
-def take_challenge(music_file, typed_lyrics, key, language):
-    st.write("Listen to music since you have to record 15seconds after that")
-    st.audio(music_file)
-    audio_value = st.experimental_audio_input("Sing Rest of music:🎙️", key=key)
-    if audio_value:
-        with open("user_sing.mp3", "wb") as f:
-            f.write(audio_value.getbuffer())
-            
-        if language == "en":
-            english_model = whisper.load_model("base.en")
-            user_lyrics = english_model.transcribe("user_sing.mp3", language=language)["text"]
-        else:
-            persian_model = Model.load("hezarai/whisper-small-fa")
-            user_lyrics = persian_model.predict("user_sing.mp3")[0]["text"]
-            
-        st.write(user_lyrics)
-        similarity_score = cosine_sim(typed_lyrics, user_lyrics)
-        if similarity_score > 0.85:
-            st.success('Awsome! You are doing great', icon="✅")
-            st.markdown('<style>div.stAlert { background-color: rgba(3, 67, 24, 0.9); }</style>', unsafe_allow_html=True)
-        else:
-            st.error('Awful! Try harder next time', icon="🚨")
-            st.markdown('<style>div.stAlert { background-color: rgba(241, 36, 36, 0.9); }</style>', unsafe_allow_html=True)
-
-def change_volume(input_file, output_file, volume_factor):
-    sound = AudioSegment.from_mp3(input_file)
-    volume_changed = sound + volume_factor
-    volume_changed.export(output_file, format="mp3")
-
-def change_speed(input_file, output_file, speed_factor):
-    sound, sr = librosa.load(input_file)
-    speed_changed = librosa.effects.time_stretch(sound, rate=speed_factor)
-    sf.write(output_file, speed_changed, sr)
-
-def change_pitch(input_file, output_file, pitch_factor):
-    sound, sr = librosa.load(input_file)
-    pitch_changed = librosa.effects.pitch_shift(sound, sr=sr, n_steps=pitch_factor)
-    sf.write(output_file, pitch_changed, sr)
-
-def low_pass_filter(input_file, output_file, cutoff_freq):
-    sound = AudioSegment.from_mp3(input_file)
-    low_filtered_sound = sound.low_pass_filter(cutoff_freq)
-    low_filtered_sound.export(output_file, format="mp3")
-
-def high_pass_filter(input_file, output_file, cutoff_freq):
-    sound = AudioSegment.from_mp3(input_file)
-    high_filtered_sound = sound.high_pass_filter(cutoff_freq)
-    high_filtered_sound.export(output_file, format="mp3")
-
-def pan_left_right(input_file, output_file, pan_factor):
-    sound = AudioSegment.from_mp3(input_file)
-    pan_sound = sound.pan(pan_factor)
-    pan_sound.export(output_file, format="mp3")
-
-def fade_in_ms(input_file, output_file, fade_factor):
-    sound = AudioSegment.from_mp3(input_file)
-    faded_sound = sound.fade_in(fade_factor)
-    faded_sound.export(output_file, format="mp3")
-
-def fade_out_ms(input_file, output_file, fade_factor):
-    sound = AudioSegment.from_mp3(input_file)
-    faded_sound = sound.fade_out(fade_factor)
-    faded_sound.export(output_file, format="mp3")
-
-
-            
+
 tab1, tab2 = st.tabs(["Take Challenge", "Make Challenge"])
 
 with tab1:
@@ -223,8 +151,13 @@ with tab2:
             fade_out_ms(current_input, output_file, fade_out_time)
             current_input = output_file
 
+        music_background_checkbox = st.checkbox("Keep Music Background?")
 
         st.write("Now type what user should sing:")
         typed_lyrics = st.text_area("Lyrics to be singed:")
         st.write('------')
-        take_challenge(current_input, typed_lyrics, "unique_key_1", language)
+        if music_background_checkbox:
+            result_list = split_vocals(current_input)
+            take_challenge(current_input, typed_lyrics, "unique_key_1", language, music_background_checkbox, result_list[0])
+        else:
+            take_challenge(current_input, typed_lyrics, "unique_key_1", language, music_background_checkbox)