app.py

import argparse
import binascii
import glob
import os
import os.path
import numpy as np
import matplotlib.pyplot as plt
import random
import sys
import tempfile
import time
import torch
from PIL import Image
from diffusers import StableDiffusionPipeline

import gradio as gr

import artist_lib

from dotenv import load_dotenv
load_dotenv()
SERVER_NAME = os.getenv("SERVER_NAME")

drawdemo = gr.Interface(
    fn=artist_lib.draw,
    inputs=[
        gr.Text(label="Drawing description text", value="hindu mandala neon orange and blue"),
        gr.Dropdown(label='Model', choices=["stable-diffusion-2", "stable-diffusion-2-1", "stable-diffusion-v1-5"], value="stable-diffusion-v1-5"),
        gr.Checkbox(label="Force-New"),
    ],
    outputs="image",
        examples=[
        ['van gogh dogs playing poker', "stable-diffusion-v1-5", False],
        ['picasso the scream', "stable-diffusion-v1-5", False],
        ['dali american gothic', "stable-diffusion-v1-5", False],
        ['matisse mona lisa', "stable-diffusion-v1-5", False],
        ['maxfield parrish angel in lake ', "stable-diffusion-v1-5", False],
        ['peter max dogs playing poker', "stable-diffusion-v1-5", False],
        ['hindu mandala copper and patina green', "stable-diffusion-v1-5", False],
        ['hindu mandala fruit salad', "stable-diffusion-v1-5", False],
        ['hindu mandala neon green black and purple', "stable-diffusion-v1-5", False],
        ['astronaut riding a horse on mars', "stable-diffusion-v1-5", False]
    ],
)

AudioDemo = gr.Interface(
    fn=artist_lib.generate_tone,
    inputs=[
        gr.Dropdown(artist_lib.notes, type="index"),
        gr.Slider(4, 6, step=1),
        gr.Textbox(value=1, label="Duration in seconds")
    ],
    outputs="audio"
)

imageClassifierDemo = gr.Interface(
    fn=artist_lib.imageClassifier,
    inputs="image",
    outputs="text"
)

audioGeneratorDemo = gr.Interface(
    fn=artist_lib.audioGenerator,
    inputs="text",
    outputs="audio",
    examples=[
        ['balsamic beats'],
        ['dance the night away']
    ]
)

nameMyPetDemo = gr.Interface(
    fn=artist_lib.nameMyPet,
    inputs=[
        gr.Text(label="What type of animal is your pet?", value="green cat")
    ],
    outputs="text",
        examples=[
        ['dog'],
        ['pink dolphin'],
        ['elevated elephant'],
        ['green monkey'],
        ['bionic beaver'],
        ['felonous fish'],
        ['delinquent dog'],
        ['dragging donkey'],
        ['stinky skunk'],
        ['pink unicorn'],
        ['naughty narwahl'],
        ['blue cat']
    ],
)

blog_writer_demo = gr.Interface(
    fn=artist_lib.write_blog,
    inputs=[
        gr.Text(label="Blog description text", value="machine learning can be used to track chickens"),
        gr.Dropdown(label='Model', choices=["gpt-neo-1.3B", "gpt-neo-2.7B"], value="gpt-neo-1.3B"),
        gr.Number(label='Minimum word count', value=50, precision=0),
        gr.Number(label='Maximum word count', value=50, precision=0),
        gr.Checkbox(label="Force-New"),
    ],
    outputs="text",
        examples=[
        ['machine learning can be used to track chickens', "gpt-neo-1.3B", 50, 50, False],
        ['music and machine learning', "gpt-neo-2.7B", 50, 50, False]
    ],
)

generateAudioDemo = gr.Interface(
    fn=artist_lib.generate_spectrogram_audio_and_loop,
    title="Audio Diffusion",
    description="Generate audio using Huggingface diffusers.\
        The models without 'latent' or 'ddim' give better results but take about \
            20 minutes without a GPU. For GPU, you can use \
                [colab](https://colab.research.google.com/github/teticio/audio-diffusion/blob/master/notebooks/gradio_app.ipynb) \
                    to run this app.",
    inputs=[
        gr.Dropdown(label="Model",
                    choices=[
                        "teticio/audio-diffusion-256",
                        "teticio/audio-diffusion-breaks-256",
                        "teticio/audio-diffusion-instrumental-hiphop-256",
                        "teticio/audio-diffusion-ddim-256",
                        "teticio/latent-audio-diffusion-256",
                        "teticio/latent-audio-diffusion-ddim-256"
                    ],
                    value="teticio/latent-audio-diffusion-ddim-256")
    ],
    outputs=[
        gr.Image(label="Mel spectrogram", image_mode="L"),
        gr.Audio(label="Audio"),
        gr.Audio(label="Loop"),
    ],
    allow_flagging="never")

with gr.Blocks() as gallerydemo:
    with gr.Column(variant="panel"):
        with gr.Row(variant="compact"):
            text = gr.Textbox(
                label="Enter your prompt",
                show_label=False,
                max_lines=1,
                placeholder="Enter your prompt"
            )
            btn = gr.Button("Generate image")

        gallery = gr.Gallery(
            label="Generated images", show_label=False, elem_id="gallery"
        )

    btn.click(artist_lib.fake_gan, None, gallery)

#artist = gr.TabbedInterface( [drawdemo, blog_writer_demo, gallerydemo], ["Draw", "Bloggr", "Gallery"])
#artist = gr.TabbedInterface( [drawdemo, blog_writer_demo, imageClassifierDemo, generateAudioDemo, audioGeneratorDemo, AudioDemo, nameMyPetDemo], ["Draw", "Bloggr", "imageClassifier", "generateAudio", "audioGenerator", "AudioDemo", "nameMyPet"])
artist = gr.TabbedInterface( [drawdemo, imageClassifierDemo, generateAudioDemo, nameMyPetDemo, blog_writer_demo], ["Draw", "imageClassifier", "generateAudio", "nameMyPet", "Bloggr"])

artist.queue(
    max_size = 4
)
artist.launch()