Update app.py

app.py

@@ -1,40 +1,89 @@
+import os
 import gradio as gr
+from pathlib import Path
+import torch
+from transformers import pipeline
+
+# --- Import your OOP modules ---
 from translation import Translator, CONFIG
 from tts_engine import TTSEngine
 
-translator = Translator(n2n_enabled=CONFIG["features"]["nigerian_to_nigerian_enabled"])
-tts_engine = TTSEngine(use_coqui=False)  # default: pyttsx3
+# --- Init core objects ---
+translator = Translator(
+    n2n_enabled=CONFIG["features"]["nigerian_to_nigerian_enabled"]
+)
+tts_engine = TTSEngine(use_coqui=False)  # default to pyttsx3
+
+# --- Whisper STT pipeline (CPU safe) ---
+device = 0 if torch.cuda.is_available() else -1
+stt_pipeline = pipeline("automatic-speech-recognition", model="openai/whisper-small", device=device)
 
+# --- Core translate + TTS function ---
 def translate_and_speak(input_text, input_lang, output_lang, use_coqui, clone_voice):
-    # Translate
+    if not input_text:
+        return "⚠️ No input detected", None
+
+    # 1. Translate
     translated = translator.translate(input_text, input_lang, output_lang)
 
-    # TTS
+    # 2. TTS
     tts_engine.use_coqui = use_coqui
-    audio_path = tts_engine.speak(translated, lang=output_lang, voice_clone=clone_voice)
+    audio_path = tts_engine.speak(
+        translated,
+        lang=output_lang,
+        voice_clone=clone_voice,
+    )
 
     return translated, audio_path
 
-
+# --- Gradio App ---
 with gr.Blocks() as demo:
     gr.Markdown("## 🌍 Nigerian Voice Translator Assistant")
 
     with gr.Row():
-        input_lang = gr.Dropdown(choices=["yoruba","hausa","igbo","pidgin","esan","tiv","calabar","benin"], value="yoruba", label="Input Language")
-        output_lang = gr.Dropdown(choices=["english","yoruba","hausa","igbo","pidgin","esan","tiv","calabar","benin"], value="english", label="Output Language")
+        input_lang = gr.Dropdown(
+            choices=["yoruba", "hausa", "igbo", "pidgin", "esan", "tiv", "calabar", "benin"],
+            value="yoruba",
+            label="Input Language"
+        )
+        output_lang = gr.Dropdown(
+            choices=["english", "yoruba", "hausa", "igbo", "pidgin", "esan", "tiv", "calabar", "benin"],
+            value="english",
+            label="Output Language"
+        )
+
+    with gr.Tab("🎙️ Voice Input"):
+        mic_input = gr.Audio(sources=["microphone"], type="filepath", label="Speak here")
+
+    with gr.Tab("⌨️ Text Input"):
+        text_input = gr.Textbox(label="Enter text")
 
-    input_text = gr.Textbox(label="Enter text")
     use_coqui = gr.Checkbox(label="Use Coqui TTS (natural accents)", value=False)
     clone_voice = gr.Checkbox(label="Clone my voice if available", value=False)
 
     translate_btn = gr.Button("Translate & Speak")
+
     output_text = gr.Textbox(label="Translation")
     output_audio = gr.Audio(label="Spoken Output", type="filepath")
 
+    # --- Events ---
+    def handle_input(mic_input, text_input, input_lang, output_lang, use_coqui, clone_voice):
+        # If mic audio provided → transcribe with Whisper
+        if mic_input:
+            print(f"🎙️ Running Whisper STT on {mic_input}")
+            result = stt_pipeline(mic_input)
+            input_text = result["text"].strip()
+        else:
+            input_text = text_input.strip() if text_input else ""
+
+        return translate_and_speak(input_text, input_lang, output_lang, use_coqui, clone_voice)
+
     translate_btn.click(
-        fn=translate_and_speak,
-        inputs=[input_text, input_lang, output_lang, use_coqui, clone_voice],
+        fn=handle_input,
+        inputs=[mic_input, text_input, input_lang, output_lang, use_coqui, clone_voice],
         outputs=[output_text, output_audio]
     )
 
-demo.launch()
+# --- Run ---
+if __name__ == "__main__":
+    demo.launch()