diff --git a/transcribe.py b/transcribe.py
index d501687..f8c7d42 100644
--- a/transcribe.py
+++ b/transcribe.py
@@ -6,7 +6,6 @@ try:
     import lzma
 except ImportError:
     mock_lzma = MagicMock()
-    # Add common constants that libraries expect from lzma
     mock_lzma.FORMAT_XZ = 1
     mock_lzma.FORMAT_ALONE = 2
     mock_lzma.FORMAT_RAW = 3
@@ -27,7 +26,13 @@ from transformers import MarianMTModel, MarianTokenizer
 
 # Parameters
 WHISPER_MODEL = "mlx-community/whisper-small.en-mlx"
-TRANSLATE_MODEL = "Helsinki-NLP/opus-mt-en-es"
+# List of language pairs (English to ...)
+TARGET_LANGS = {
+    "Spanish": "Helsinki-NLP/opus-mt-en-es",
+    "French":  "Helsinki-NLP/opus-mt-en-fr",
+    "Arabic":  "Helsinki-NLP/opus-mt-en-ar"
+}
+
 CHANNELS = 1
 SAMPLERATE = 16000
 BLOCK_SIZE = 512
@@ -43,22 +48,26 @@ def callback(indata, frames, time, status):
     audio_queue.put(indata.copy())
 
 def main():
-    # Set device for translation model
     device = "mps" if torch.backends.mps.is_available() else "cpu"
-    print(f"Using device for translation: {device}")
+    print(f"Using device: {device}")
 
     print(f"Loading Whisper model '{WHISPER_MODEL}'...")
     
-    print(f"Loading dedicated translation model '{TRANSLATE_MODEL}'...")
-    tokenizer = MarianTokenizer.from_pretrained(TRANSLATE_MODEL)
-    model = MarianMTModel.from_pretrained(TRANSLATE_MODEL).to(device)
+    # Dictionary to hold models and tokenizers
+    translation_engines = {}
+    
+    for lang_name, model_id in TARGET_LANGS.items():
+        print(f"Loading {lang_name} translation model ({model_id})...")
+        tokenizer = MarianTokenizer.from_pretrained(model_id)
+        model = MarianMTModel.from_pretrained(model_id).to(device)
+        translation_engines[lang_name] = (model, tokenizer)
 
     print("Loading Silero VAD model...")
     vad_model = load_silero_vad()
     
     print("Models loaded.")
 
-    print(f"\nStarting live transcription & translation... (Press Ctrl+C to stop)")
+    print(f"\nStarting live transcription & multiple translations... (Press Ctrl+C to stop)")
     
     audio_buffer = []
     speech_started = False
@@ -89,19 +98,20 @@ def main():
                         
                         if (buffer_len_samples - last_end) > (SAMPLERATE * MIN_SILENCE_DURATION_MS / 1000) or buffer_len_samples > BUFFER_LIMIT:
                             
-                            # 1. Transcribe (English)
+                            # 1. Transcribe once
                             result = mlx_whisper.transcribe(current_audio, path_or_hf_repo=WHISPER_MODEL)
                             original_text = result['text'].strip()
                             
                             if original_text:
-                                # 2. Translate
-                                inputs = tokenizer(original_text, return_tensors="pt", padding=True).to(device)
-                                with torch.no_grad():
-                                    translated_tokens = model.generate(**inputs)
-                                translated_text = tokenizer.decode(translated_tokens[0], skip_special_tokens=True)
+                                print(f"\n[EN]: {original_text}")
                                 
-                                print(f"\nEN: {original_text}")
-                                print(f"ES: {translated_text}")
+                                # 2. Translate to all targets
+                                for lang_name, (model, tokenizer) in translation_engines.items():
+                                    inputs = tokenizer(original_text, return_tensors="pt", padding=True).to(device)
+                                    with torch.no_grad():
+                                        translated_tokens = model.generate(**inputs)
+                                    translated_text = tokenizer.decode(translated_tokens[0], skip_special_tokens=True)
+                                    print(f"[{lang_name[:2].upper()}]: {translated_text}")
                             
                             audio_buffer = []
                             speech_started = False