Debugging

app.py

@@ -7,7 +7,7 @@ import requests
 from datetime import datetime
 import tempfile
 from gtts import gTTS
-import os
+
 
 DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
 
@@ -45,7 +45,7 @@ def save_to_supabase(input_text, output_text, direction):
         response = requests.post(f"{SUPABASE_URL}/rest/v1/{table}", json=payload, headers=headers)
         return "✅ Saved successfully!" if response.status_code == 201 else "❌ Failed to save."
     except Exception as e:
-        print("Save error:", e)
+        logging.error("Save error: %s", e)
         return "❌ Save error."
 
 def get_translation_history(direction):
@@ -62,11 +62,11 @@ def get_translation_history(direction):
             return "\n\n".join([f"Input: {r['input_text']} → Output: {r['output_text']}" for r in data])
         return "Failed to load history."
     except Exception as e:
-        print("History error:", e)
+        logging.error("History error: %s", e)
         return "Error loading history."
 
-# --- Translation ---
-def translate(text, direction):
+# --- Translation with TTS integration ---
+def translate(text, direction, generate_tts=False):
     if not text.strip():
         return "Enter some text.", None
 
@@ -84,9 +84,15 @@ def translate(text, direction):
             output = model.generate(**tokens, max_length=256, num_beams=5)
         result = tokenizer.batch_decode(output, skip_special_tokens=True)
         final = ip.postprocess_batch(result, lang=tgt_lang)[0]
-        return final, None
+        
+        # Generate TTS for KS→EN direction if requested
+        audio_path = None
+        if generate_tts and direction == "ks_to_en":
+            audio_path = synthesize_tts(final)
+            
+        return final, audio_path
     except Exception as e:
-        print("Translation error:", e)
+        logging.error("Translation error: %s", e)
         return "⚠️ Translation failed.", None
 
 # --- TTS for English output ---
@@ -97,28 +103,36 @@ def synthesize_tts(text):
             tts.save(f.name)
             return f.name
     except Exception as e:
-        print("TTS error:", e)
+        logging.error("TTS error: %s", e)
         return None
 
-# --- STT only for en_to_ks ---
-def generate_stt_for_input(audio_path, direction):
-    if direction != "en_to_ks":
-        return "⚠️ Audio input is only supported for English to Kashmiri.", "", None
-
+# --- STT for English audio ---
+def transcribe_audio(audio_path):
     try:
+        if not audio_path:
+            return None, "No audio file provided"
+            
+        # Convert to mono if needed
+        if isinstance(audio_path, tuple):
+            audio_path = audio_path[1]  # Gradio returns (sample_rate, audio_data)
+            
         transcription = asr(audio_path)["text"]
+        return transcription, None
     except Exception as e:
-        print("STT error:", e)
-        return "⚠️ Transcription failed.", "", None
+        logging.error("STT error: %s", e)
+        return None, f"⚠️ Transcription failed: {str(e)}"
 
-    translated, _ = translate(transcription, direction)
-    return transcription, translated, None
+# --- Handle audio translation ---
+def handle_audio_translation(audio_path, direction):
+    if direction != "en_to_ks":
+        return "⚠️ Audio input is only supported for English to Kashmiri.", "", None
 
-# --- Generate TTS for English output ---
-def generate_tts_for_output(output_text, direction):
-    if direction == "ks_to_en" and output_text.strip():
-        return synthesize_tts(output_text)
-    return None
+    transcription, error = transcribe_audio(audio_path)
+    if error:
+        return error, "", None
+        
+    translated, _ = translate(transcription, direction)
+    return "", transcription, translated
 
 # --- Switch UI direction ---
 def switch_direction(direction, input_text_val, output_text_val):
@@ -145,23 +159,24 @@ with gr.Blocks() as interface:
     history = gr.Textbox(label="Translation History", lines=8, interactive=False)
 
     with gr.Row():
-        audio_input = gr.Audio(type="filepath", label="🎙️ Record English audio")
+        audio_input = gr.Audio(type="filepath", label="🎙️ Record English audio", sources=["microphone"])
         audio_output = gr.Audio(label="🔊 English TTS", interactive=False)
 
-    stt_button = gr.Button("🎤 Transcribe & Translate (EN → KS Only)")
-    tts_button = gr.Button("🔊 Generate English Speech (KS → EN Only)")
+    with gr.Row():
+        stt_button = gr.Button("🎤 Transcribe & Translate (EN → KS)")
+        tts_button = gr.Button("🔊 Translate & Speak (KS → EN)")
 
     # Events
     translate_button.click(
         fn=translate,
-        inputs=[input_text, translation_direction],
+        inputs=[input_text, translation_direction, gr.State(False)],
         outputs=[output_text, audio_output]
     )
     
     tts_button.click(
-        fn=generate_tts_for_output,
-        inputs=[output_text, translation_direction],
-        outputs=audio_output
+        fn=translate,
+        inputs=[input_text, translation_direction, gr.State(True)],
+        outputs=[output_text, audio_output]
     )
 
     save_button.click(
@@ -181,9 +196,9 @@ with gr.Blocks() as interface:
     )
 
     stt_button.click(
-        fn=generate_stt_for_input,
+        fn=handle_audio_translation,
         inputs=[audio_input, translation_direction],
-        outputs=[input_text, output_text, audio_output]
+        outputs=[save_status, input_text, output_text]
     )
 
 if __name__ == "__main__":