app.py

#from transformers import pipeline
import gradio as gr
#import torch
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.float16)
image_model = image_model.to("cuda")

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=["text", "image"],
).launch()