Create app.py

app.py

@@ -0,0 +1,47 @@
+import streamlit as st
+from transformers import MarianMTModel, MarianTokenizer
+import whisper
+import tempfile
+import base64
+import os
+from audiorecorder import audiorecorder  # pip install streamlit-audiorecorder
+
+# Load models once
+@st.cache_resource
+def load_models():
+    whisper_model = whisper.load_model("base")
+    tokenizer = MarianTokenizer.from_pretrained("Helsinki-NLP/opus-mt-ar-en")
+    translator = MarianMTModel.from_pretrained("Helsinki-NLP/opus-mt-ar-en")
+    return whisper_model, tokenizer, translator
+
+st.title("🎙️ Live Arabic Sermon Translator")
+
+st.markdown("Click the mic, say something in Arabic, and wait a few seconds for translation.")
+
+# Record audio
+audio = audiorecorder("Start Recording", "Stop Recording")
+
+if len(audio) > 0:
+    st.audio(audio.tobytes(), format="audio/wav")
+
+    # Save audio to temp file
+    with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as f:
+        f.write(audio.tobytes())
+        temp_wav_path = f.name
+
+    st.info("Transcribing Arabic...")
+    whisper_model, tokenizer, translator = load_models()
+    transcription = whisper_model.transcribe(temp_wav_path, language="ar")
+    arabic_text = transcription["text"]
+    st.markdown("### Arabic")
+    st.write(arabic_text)
+
+    st.info("Translating to English...")
+    tokens = tokenizer(arabic_text, return_tensors="pt", padding=True)
+    output = translator.generate(**tokens)
+    english_text = tokenizer.decode(output[0], skip_special_tokens=True)
+
+    st.markdown("### English")
+    st.success(english_text)
+
+    os.remove(temp_wav_path)