File size: 2,413 Bytes
7f3a4c0 20200de 7f3a4c0 2e7454b 7f3a4c0 f93771f b632575 7f3a4c0 0a97cd1 7f3a4c0 0a97cd1 f93771f |
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 |
import gradio as gr
import os, subprocess, torchaudio
import torch
from PIL import Image
block = gr.Blocks()
def pad_image(image):
w, h = image.size
if w == h:
return image
elif w > h:
new_image = Image.new(image.mode, (w, w), (0, 0, 0))
new_image.paste(image, (0, (w - h) // 2))
return new_image
else:
new_image = Image.new(image.mode, (h, h), (0, 0, 0))
new_image.paste(image, ((h - w) // 2, 0))
return new_image
def calculate(image_in, audio_in):
waveform, sample_rate = torchaudio.load(audio_in)
waveform = torch.mean(waveform, dim=0, keepdim=True)
torchaudio.save("/content/audio.wav", waveform, sample_rate, encoding="PCM_S", bits_per_sample=16)
image = Image.open(image_in)
image = pad_image(image)
image.save("image.png")
pocketsphinx_run = subprocess.run(['pocketsphinx', '-phone_align', 'yes', 'single', '/content/audio.wav'], check=True, capture_output=True)
jq_run = subprocess.run(['jq', '[.w[]|{word: (.t | ascii_upcase | sub("<S>"; "sil") | sub("<SIL>"; "sil") | sub("\\\(2\\\)"; "") | sub("\\\(3\\\)"; "") | sub("\\\(4\\\)"; "") | sub("\\\[SPEECH\\\]"; "SIL") | sub("\\\[NOISE\\\]"; "SIL")), phones: [.w[]|{ph: .t | sub("\\\+SPN\\\+"; "SIL") | sub("\\\+NSN\\\+"; "SIL"), bg: (.b*100)|floor, ed: (.b*100+.d*100)|floor}]}]'], input=pocketsphinx_run.stdout, capture_output=True)
with open("test.json", "w") as f:
f.write(jq_run.stdout.decode('utf-8').strip())
# device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
os.system(f"cd /content/one-shot-talking-face && python3 -B test_script.py --img_path /content/image.png --audio_path /content/audio.wav --phoneme_path /content/test.json --save_dir /content/train")
return "/content/train/image_audio.mp4"
def run():
with block:
with gr.Group():
with gr.Box():
with gr.Row().style(equal_height=True):
image_in = gr.Image(show_label=False, type="filepath")
audio_in = gr.Audio(show_label=False, type='filepath')
video_out = gr.Video(show_label=False)
with gr.Row().style(equal_height=True):
btn = gr.Button("Generate")
btn.click(calculate, inputs=[image_in, audio_in], outputs=[video_out])
block.queue()
block.launch(server_name="0.0.0.0", server_port=7860)
if __name__ == "__main__":
run()
|