app.py

##!/usr/bin/python3
# -*- coding: utf-8 -*-
import os

print("Installing...")
os.system("pip install gradio")
os.system("pip install tf-keras")
os.system("pip install diffusers")
os.system("pip install accelerate")
os.system("pip install transformers")
os.system("pip install numpy")
os.system("pip install torch")
#os.system("pip install --upgrade pip")

print("Installing Finished!")

##!/usr/bin/python3
# -*- coding: utf-8 -*-

from transformers import pipeline
import gradio as gr
import os
import torch
import accelerate
from diffusers import StableDiffusionPipeline, EulerDiscreteScheduler


model_id = "stabilityai/stable-diffusion-2"

scheduler = EulerDiscreteScheduler.from_pretrained(model_id, subfolder="scheduler")
image_model = StableDiffusionPipeline.from_pretrained(model_id, scheduler=scheduler, torch_dtype=torch.float32)
image_model = image_model.to("cpu")

model = pipeline("automatic-speech-recognition","facebook/wav2vec2-large-xlsr-53-spanish")

def transcribe_text_audio(mic=None, file=None):
    if mic is not None:
        audio = mic
    elif file is not None:
        audio = file
    else:
        return "No se ha detectado ninguna entrada de audio"

    transcription = model(audio)["text"]

    image = image_model(transcription).images[0]

    image = image.convert("RGB")
    return transcription, image


gr.Interface(
    fn=transcribe_text_audio,
    inputs=[
        gr.Audio(sources=["microphone"], type="filepath"),
        gr.Audio(sources=["upload"], type="filepath"),
    ],
    outputs=[
      gr.Textbox(label="Transcripción del Audio"),
      gr.Image(label="Imagen Generada")
    ],
    title="[ESpañol] - Audio -> Texto -> Imagen",
    description="Esta aplicación transcribe el audio a texto para convertirlo en una imagen descriptiva."
).launch()