Create app.py

app.py

@@ -0,0 +1,63 @@
+
+from datasets import load_dataset
+from transformers import pipeline
+import soundfile as sf
+import torch
+import gradio as gr
+import numpy as np
+
+def predict_image(image):
+  classifier=pipeline(task='image-classification' ,model='google/vit-base-patch16-224')
+  prediction=classifier(image)
+  result=prediction[0]['label']
+  return result
+
+def translate_to_arabic(text):
+    translator=pipeline(task='translation_en_to_ar' ,model='Helsinki-NLP/opus-mt-en-ar')
+    result=translator(text,max_length=100)
+    return result[0]['translation_text']
+
+def text_to_speech(text):
+  model=pipeline(task='text-to-speech', model='MBZUAI/speecht5_tts_clartts_ar')
+  embedding_dataset=load_dataset('herwoww/arabic_xvector_embeddings',split='validation')
+  speaker_embedding=torch.tensor(embedding_dataset[100]['speaker_embeddings']).unsqueeze(0) ##It becomes a 2-D tensor:
+  speech=model(text,forward_params={'speaker_embeddings':speaker_embedding})
+
+  return (speech['sampling_rate'],np.array(speech['audio'],dtype=np.float32))
+
+with gr.Blocks() as demo:
+  gr.Markdown("## Image classification, Arabic Translation and text to speech interface")
+
+  with gr.Row():
+    with gr.Column():
+      image_input=gr.Image(type="pil",label='upload_image')
+      predict_button=gr.Button('predict')
+      Prediction_Result=gr.Textbox(label='Prediction Result')
+
+    predict_button.click(
+      fn=predict_image,
+      inputs=image_input,
+      outputs=Prediction_Result
+    )
+  with gr.Row():
+    with gr.Column():
+      translated_text=gr.Textbox(label='Translated Text')
+      translated_button=gr.Button('Translate To Arabic')
+
+      translated_button.click(
+        fn=translate_to_arabic,
+        inputs=Prediction_Result,
+        outputs=translated_text
+      )
+
+  with gr.Row():
+      to_speech_button=gr.Button('convert To Audio')
+      audio_output=gr.Audio(label='Audio Output')
+
+      to_speech_button.click(
+        fn=text_to_speech,
+        inputs=translated_text,
+        outputs=audio_output
+      )
+
+demo.launch(share=True)