Helsinki-NLP

app.py

@@ -1,14 +1,20 @@
 import gradio as gr
 import librosa
 import torch
-from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor, MBartForConditionalGeneration, MBart50Tokenizer
+from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor, MBartForConditionalGeneration, MBart50Tokenizer, MarianMTModel, MarianTokenizer
 
 # Load pre-trained models
 model = Wav2Vec2ForCTC.from_pretrained("boumehdi/wav2vec2-large-xlsr-moroccan-darija")
 processor = Wav2Vec2Processor.from_pretrained("boumehdi/wav2vec2-large-xlsr-moroccan-darija")
 
-translation_model = MBartForConditionalGeneration.from_pretrained("facebook/mbart-large-50-many-to-many-mmt")
-translation_tokenizer = MBart50Tokenizer.from_pretrained("facebook/mbart-large-50-many-to-many-mmt", src_lang="ar_AR")
+#translation_model = MBartForConditionalGeneration.from_pretrained("facebook/mbart-large-50-many-to-many-mmt")
+#translation_tokenizer = MBart50Tokenizer.from_pretrained("facebook/mbart-large-50-many-to-many-mmt", src_lang="ar_AR")
+
+# Charger le modèle de traduction Arabe -> Anglais
+translation_model_name = "Helsinki-NLP/opus-mt-ar-en"
+translation_model = MarianMTModel.from_pretrained(translation_model_name)
+translation_tokenizer = MarianTokenizer.from_pretrained(translation_model_name)
+
 
 def transcribe_audio(audio):
     audio_array, sr = librosa.load(audio, sr=16000)