Update app.py

app.py

@@ -1,9 +1,10 @@
+from faster_whisper import WhisperModel
 import streamlit as st
-import ctranslate2
-import transformers
-from translator import translate
-st.title("On Device Translator")
-user_input = st.text_area("Enter text to be translated in English here.") 
+from translator import translate  
+model_size = "small"
+model = WhisperModel(model_size, device="cpu", compute_type="int8")
+st.title("Audio to Text Translator")
+audio_file = st.file_uploader("Upload an audio file", type=["mp3", "wav", "ogg", "flac"])
 languages = {
     "English": "eng_Latn",
     "Japanese": "jpn_Jpan",
@@ -16,7 +17,30 @@ languages = {
     "Russian": "rus_Cyrl"
 }
 selected_language = st.selectbox("Select Target Language", list(languages.keys()))
-if st.button("Translate"):
-    translated_text = translate(user_input, tgt_lang=languages[selected_language])
-    st.write(f"Translated text: {translated_text}")
+if st.button("Transcribe and Translate"):
+    if audio_file is not None:
+        with open("temp_audio.mp3", "wb") as f:
+            f.write(audio_file.getbuffer())
+        segments, info = model.transcribe("temp_audio.mp3", beam_size=5)
+        st.write("Detected language '%s' with probability %f" % (info.language, info.language_probability))
+        col1, col2 = st.columns(2)
+
+        with col1:
+            st.subheader("Transcription")
+
+        with col2:
+            st.subheader("Translation")
+
+        # Process each segment individually
+        for segment in segments:
+            # Translate the segment text
+            translated_text = translate(segment.text, tgt_lang=languages[selected_language])
+            
+            with col1:
+                st.markdown(f"**[{segment.start:.2f}s - {segment.end:.2f}s]** {segment.text}")
+            
+            with col2:
+                st.markdown(f"**[{segment.start:.2f}s - {segment.end:.2f}s]** {translated_text}")
+    else:
+        st.error("Please upload an audio file.")