Update app.py

app.py

@@ -152,7 +152,7 @@ def translate_local(text_to_translate, target_lang='ta-IN', device=None):
         chunks.append(current_chunk.strip())
 
     # 4. Translate each chunk and combine the results
-    translator = MyMemoryTranslator(source='en-GB', target="ta-IN")
+    translator = MyMemoryTranslator(source='en-GB', target=target_lang)
     translated_chunks = []
     for chunk in chunks:
         try:
@@ -177,7 +177,7 @@ def synthesize_speech(synth_text, target_lang, gender="Male", pace="normal", out
     print("> Loading IndicF5 TTS model (ai4bharat/IndicF5)...")
     indicf5_repo_id = "ai4bharat/IndicF5"
     token = os.environ.get("HF_TOKEN")
-    tts_model = AutoModel.from_pretrained(indicf5_repo_id, token=token, trust_remote_code=True).to(device)
+    tts_model = AutoModel.from_pretrained(indicf5_repo_id, token=token, trust_remote_code=True, device_map=device)
     
     audio = tts_model(synth_text, ref_audio_path=ref_audio_path, ref_text=ref_text)
 
@@ -260,8 +260,8 @@ def audio_pipeline_run(audio_path,target_lang):
     original_text , pace = transcribe_audio(audio_path)
     translated_text = translate_local(original_text,target_lang)
     translated_audio = synthesize_speech(translated_text, target_lang, "Male", pace)
-    synced_translated_audio = match_audio_duration(original_audio_file, translated_audio) 
-    cloned_synced_translated_audio = clone_voice(original_audio_file, synced_translated_audio, target_lang, "Male", pace)
+    synced_translated_audio = match_audio_duration(audio_path, translated_audio) 
+    cloned_synced_translated_audio = clone_voice(audio_path, synced_translated_audio, target_lang, "Male", pace)
     return cloned_synced_translated_audio