Update app.py

app.py

@@ -3,22 +3,29 @@ import gradio as gr
 from pathlib import Path
 import torch
 from transformers import pipeline
+from langdetect import detect
 
-# --- Import your OOP modules ---
+# --- Import OOP modules ---
 from translation import Translator, CONFIG
 from tts_engine import TTSEngine
 
-# --- Init core objects ---
+# --- Init core ---
 translator = Translator(
     n2n_enabled=CONFIG["features"]["nigerian_to_nigerian_enabled"]
 )
-tts_engine = TTSEngine(use_coqui=False)  # default to pyttsx3
 
-# --- Whisper STT pipeline (CPU safe) ---
+# On Hugging Face: disable pyttsx3 (no espeak). Default = Coqui
+USE_PYTTSX3 = os.environ.get("USE_PYTTSX3", "false").lower() == "true"
+tts_engine = TTSEngine(use_coqui=not USE_PYTTSX3)
+
+# --- Whisper STT ---
 device = 0 if torch.cuda.is_available() else -1
-stt_pipeline = pipeline("automatic-speech-recognition", model="openai/whisper-small", device=device)
+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):
     if not input_text:
         return "⚠️ No input detected", None
@@ -36,18 +43,50 @@ def translate_and_speak(input_text, input_lang, output_lang, use_coqui, clone_vo
 
     return translated, audio_path
 
+
+def handle_input(mic_input, text_input, input_lang, output_lang, use_coqui, clone_voice):
+    input_text = ""
+
+    # Mic input → Whisper
+    if mic_input:
+        result = stt_pipeline(mic_input)
+        input_text = result["text"].strip()
+
+        # Auto language detect
+        try:
+            detected = detect(input_text)
+            print(f"🌍 Auto-detected: {detected}")
+            # Map detection to supported langs
+            if detected.startswith("yo"):
+                input_lang = "yoruba"
+            elif detected.startswith("ha"):
+                input_lang = "hausa"
+            elif detected.startswith("ig"):
+                input_lang = "igbo"
+            elif detected.startswith("en"):
+                input_lang = "english"
+            # else leave user selection
+        except Exception as e:
+            print("⚠️ Language detection failed:", e)
+
+    elif text_input:
+        input_text = text_input.strip()
+
+    return translate_and_speak(input_text, input_lang, output_lang, use_coqui, clone_voice)
+
+
 # --- 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"
+            choices=["auto", "yoruba", "hausa", "igbo", "pidgin", "esan", "tiv", "calabar", "benin", "english"],
+            value="auto",
+            label="Input Language (auto-detect by default)"
         )
         output_lang = gr.Dropdown(
-            choices=["english", "yoruba", "hausa", "igbo", "pidgin", "esan", "tiv", "calabar", "benin"],
+            choices=["english","yoruba","hausa","igbo","pidgin","esan","tiv","calabar","benin"],
             value="english",
             label="Output Language"
         )
@@ -58,7 +97,7 @@ with gr.Blocks() as demo:
     with gr.Tab("⌨️ Text Input"):
         text_input = gr.Textbox(label="Enter text")
 
-    use_coqui = gr.Checkbox(label="Use Coqui TTS (natural accents)", value=False)
+    use_coqui = gr.Checkbox(label="Use Coqui TTS (natural accents)", value=True)
     clone_voice = gr.Checkbox(label="Clone my voice if available", value=False)
 
     translate_btn = gr.Button("Translate & Speak")
@@ -66,24 +105,11 @@ with gr.Blocks() as demo:
     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=handle_input,
         inputs=[mic_input, text_input, input_lang, output_lang, use_coqui, clone_voice],
         outputs=[output_text, output_audio]
     )
 
-# --- Run ---
 if __name__ == "__main__":
     demo.launch()