app.py

import torch
import os
import random
import gradio as gr
from transformers import pipeline
import base64
from datasets import load_dataset
from diffusers import DiffusionPipeline
from huggingface_hub import login
import numpy as np

def guessanImage(model, image):
    imgclassifier  = pipeline("image-classification", model=model)
    if image is not None:  
        description = imgclassifier(image)
    return description

def guessanAge(model, image):
    imgclassifier  = pipeline("image-classification", model=model)
    if image is not None:  
        description = imgclassifier(image)
    return description    

def text2speech(model, text):
    if len(text) > 0:
        speechclassifier  = pipeline("text-to-speech", model=model)
        embeddings_dataset = load_dataset("Matthijs/cmu-arctic-xvectors", split="validation")
        speaker_embedding = torch.tensor(embeddings_dataset[7306]["xvector"]).unsqueeze(0)
        output = speechclassifier(text, forward_params={"speaker_embeddings": speaker_embedding})
        console.log(output, type(output) )
        return output

radio1 = gr.Radio(["microsoft/resnet-50", "google/vit-base-patch16-224", "apple/mobilevit-small"], label="Select a Classifier", info="Image Classifier")
tab1 = gr.Interface(
    fn=guessanImage,
    inputs=[radio1, gr.Image(type="pil")],
    outputs=["text"],
)

radio2 = gr.Radio(["nateraw/vit-age-classifier"], label="Select an Age Classifier", info="Age Classifier")
tab2 = gr.Interface(
    fn=guessanAge,
    inputs=[radio2, gr.Image(type="pil")],
    outputs=["text"],
)

radio3 = gr.Radio(["microsoft/speecht5_tts"], label="Select an tts", info="Age Classifier")
tab3 = gr.Interface(
    fn=text2speech,
    inputs=[radio3, "text"],
    outputs=[gr.Audio(label="Generated Speech", type="numpy")],
)

demo = gr.TabbedInterface([tab1, tab2, tab3], ["tab1", "tab2", "tab3"])
demo.launch()