Spaces:
Runtime error
Runtime error
File size: 2,369 Bytes
ed33636 f739d3e ed33636 5c6830a ed33636 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 |
import torch
import torchaudio
import audio_diffusion_attacks_forhf.src.test_encoder_attack as attack
import numpy as np
from transformers import EncodecModel
'''
Files edited:
- test_encoder_attack.py
- losses.py
- audio_signal.py; print statements only
'''
'''
#Set default parameters (if necessary)
args = 0
#Load pretrained protection model:
modelD = md.model(args)
modelD.load_state_dict(torch.load('path to pretrained weights', map_location=torch.device('cpu')), strict=True)
modelD.eval()
#Load pretrained mimickry model
modelM = mm.model(args)
modelM.load_state_dict(torch.load('path to pretrained weights', map_location=torch.device('cpu')), strict=True)
modelM.eval()
'''
#Define function to convert final audio format:
def float32_to_int16(waveform):
waveform = waveform / np.abs(waveform).max()
waveform = waveform * 32767
waveform = waveform.astype(np.int16)
waveform = waveform.ravel()
return waveform
#Define predict function:
def predict(inp):
#How to transform audio from string to tensor
waveform, sample_rate = torchaudio.load(inp)
encoders = [EncodecModel.from_pretrained("facebook/encodec_48khz")]
#Run modelD to disguise audio
with torch.no_grad():
waveformD = attack.poison_audio(waveform, sample_rate, encoders)
#Transform output audio into gradio-readable format
waveform = waveform.numpy()
waveform = float32_to_int16(waveform)
'''waveformD = waveformD.numpy()
waveformD = float32_to_int16(waveformD)
waveformM = waveformM.numpy()
waveformM = float32_to_int16(waveformM)
waveformDM = waveformDM.numpy()
waveformDM = float32_to_int16(waveformDM)'''
#return (sample_rate, waveformD), (sample_rate, waveformM), (sample_rate, waveformDM)
return (sample_rate, waveform), (sample_rate, waveform), (sample_rate, waveform)
#Set up gradio interface
import gradio as gr
interface = gr.Interface(
fn=predict,
inputs=gr.Audio(type="filepath"),
outputs=[gr.Audio(), gr.Audio(), gr.Audio()],
title="Music Protection Net",
description="This model is designed to add perturbations to a musical clip so that musical cloning models fail to properly reproduce the song. \n \n 1) Upload (or record) an audio file of your music. \n 2) Click submit to run the model. \n 3) Listen to and download your protected audio.",
)
interface.launch() |