app.py

# -*- coding: utf-8 -*-
"""Image & Music Generator.ipynb

Automatically generated by Colaboratory.

Original file is located at
    https://colab.research.google.com/drive/1TBH0E9wF716_A8o_yT68mSp_IFnb14_F
"""

!nvidia-smi

pip install accelerate

!pip install -q https://github.com/camenduru/stable-diffusion-webui-colab/releases/download/0.0.15/xformers-0.0.15.dev0+189828c.d20221207-cp38-cp38-linu

! pip install -U transformers diffusers gradio ftfy pydub -q

!wget https://raw.githubusercontent.com/hmartiro/riffusion-inference/main/riffusion/audio.py

import gradio as gr

import torch
from diffusers import StableDiffusionPipeline
from audio import wav_bytes_from_spectrogram_image

model_id = "riffusion/riffusion-model-v1"
pipe = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16)

pipe = pipe.to("cuda")
#pipe.enable_xformers_memory_efficient_attention()

import random
COLORS = [
    ["#ff0000", "#00ff00"],
    ["#00ff00", "#0000ff"],
    ["#0000ff", "#ff0000"],
]

from diffusers import StableDiffusionPipeline
import torch

img_model_id = "runwayml/stable-diffusion-v1-5"
img_pipe = StableDiffusionPipeline.from_pretrained(img_model_id, torch_dtype=torch.float16, revision="fp16")
img_pipe = img_pipe.to("cuda")
#img_pipe.enable_xformers_memory_efficient_attention()

prompt = 'morning sunshine'

spectogram = pipe(prompt).images[0]
wav = wav_bytes_from_spectrogram_image(spectogram)
with open("output.wav", "wb") as f:
    f.write(wav[0].getbuffer())

def audio_gen(prompt):
    spectogram = pipe(prompt).images[0]
    wav = wav_bytes_from_spectrogram_image(spectogram)
    with open("output.wav", "wb") as f:
        f.write(wav[0].getbuffer())
    print("audio saved")
    
    return ('output.wav',)

audio_gen("lazy nights")

gr.Interface(
    audio_gen,
    inputs=[gr.Textbox(label="prompt")],
    outputs=[
        gr.Audio(type='filepath')
        
    ],
    title = 'Music Generator'
).launch(debug = True)