Update app.py

app.py

@@ -72,7 +72,7 @@ def cosine_sim(text1, text2):
     return cosine_similarity(vectors)[0, 1]
     
 
-model = whisper.load_model("base")
+model = whisper.load_model("small")
 
 
 tab1, tab2 = st.tabs(["Take Challenge", "Make Challenge"])
@@ -89,7 +89,7 @@ with tab1:
         with open("user_sing.mp3", "wb") as f:
             f.write(audio_value.getbuffer())
     
-        user_lyrics = model.transcribe("user_sing.mp3")["text"]
+        user_lyrics = model.transcribe("user_sing.mp3", language="en")["text"]
         st.write(user_lyrics)
         similarity_score = cosine_sim(lyrics, user_lyrics)
         if similarity_score > 0.85:
@@ -100,7 +100,7 @@ with tab1:
             st.markdown('<style>div.stAlert { background-color: rgba(241, 36, 36, 0.9); }</style>', unsafe_allow_html=True)
 
 
-def take_challenge(music_file, typed_lyrics, key):
+def take_challenge(music_file, typed_lyrics, key, language):
     st.write('------')
     st.write("Listen to music since you have to record 15seconds after that")
     st.audio(music_file)
@@ -108,7 +108,7 @@ def take_challenge(music_file, typed_lyrics, key):
     if audio_value:
         with open("user_sing.mp3", "wb") as f:
             f.write(audio_value.getbuffer())
-        user_lyrics = model.transcribe("user_sing.mp3")["text"]
+        user_lyrics = model.transcribe("user_sing.mp3", language=language)["text"]
         st.write(user_lyrics)
         similarity_score = cosine_sim(typed_lyrics, user_lyrics)
         if similarity_score > 0.85:
@@ -123,6 +123,9 @@ def take_challenge(music_file, typed_lyrics, key):
 with tab2:
     st.write("Upload music to make challenge:")
     uploaded_file = st.file_uploader("Choose a music file", type=["mp3", "wav"])
+    language_mapping = {"English": "en", "Persian": "fa"}
+    selected_language = st.radio("Select Language", language_mapping.keys(), horizontal=True)
+    language = language_mapping[selected_language]
     if uploaded_file is not None:
         with open("raw_music.mp3", "wb") as f:
             f.write(uploaded_file.getbuffer())
@@ -139,11 +142,11 @@ with tab2:
                 trimmed_audio.export("trimmed_music.mp3", format="mp3")
                 st.write("Now type what user should sing:")
                 typed_lyrics = st.text_area("Lyrics to be singed:")
-                take_challenge("trimmed_music.mp3", typed_lyrics, "unique_key_1")
+                take_challenge("trimmed_music.mp3", typed_lyrics, "unique_key_1", language)
             else:
                 st.error('Start Time should be smaller than End Time!', icon="❌")
                 st.markdown('<style>div.stAlert { background-color: rgba(241, 36, 36, 0.9); }</style>', unsafe_allow_html=True)
         else:
             st.write("Now type what user should sing:")
             typed_lyrics = st.text_area("Lyrics to be singed:")
-            take_challenge("raw_music.mp3", typed_lyrics, "unique_key_2")
+            take_challenge("raw_music.mp3", typed_lyrics, "unique_key_2", language)