Spaces:
Running
on
T4
Running
on
T4
Update app.py
Browse files
app.py
CHANGED
@@ -1,11 +1,7 @@
|
|
1 |
import gradio as gr
|
2 |
import torch
|
3 |
|
4 |
-
from
|
5 |
-
import numpy as np
|
6 |
-
from PIL import Image
|
7 |
-
|
8 |
-
from spectro import wav_bytes_from_spectrogram_image, spectrogram_from_waveform, image_from_spectrogram
|
9 |
|
10 |
from diffusers import StableDiffusionPipeline
|
11 |
from diffusers import StableDiffusionImg2ImgPipeline
|
@@ -15,63 +11,20 @@ from share_btn import community_icon_html, loading_icon_html, share_js
|
|
15 |
MODEL_ID = "riffusion/riffusion-model-v1"
|
16 |
pipe = StableDiffusionPipeline.from_pretrained(MODEL_ID, torch_dtype=torch.float16)
|
17 |
pipe = pipe.to("cuda")
|
18 |
-
pipe2 = StableDiffusionImg2ImgPipeline.from_pretrained(MODEL_ID, torch_dtype=torch.float16)
|
19 |
-
pipe2 = pipe2.to("cuda")
|
20 |
-
|
21 |
-
def predict(prompt, audio_input, duration):
|
22 |
-
if audio_input == None:
|
23 |
-
return classic(prompt, duration)
|
24 |
-
else:
|
25 |
-
return audio_transfer(prompt, audio_input)
|
26 |
-
|
27 |
|
28 |
-
|
|
|
29 |
if duration == 5:
|
30 |
width_duration=512
|
31 |
else :
|
32 |
width_duration = 512 + ((int(duration)-5) * 128)
|
33 |
-
spec = pipe(prompt, height=512, width=width_duration).images[0]
|
34 |
print(spec)
|
35 |
wav = wav_bytes_from_spectrogram_image(spec)
|
36 |
with open("output.wav", "wb") as f:
|
37 |
f.write(wav[0].getbuffer())
|
38 |
return spec, 'output.wav', gr.update(visible=True), gr.update(visible=True), gr.update(visible=True)
|
39 |
|
40 |
-
def audio_transfer(prompt, audio):
|
41 |
-
# read uploaded file to wav
|
42 |
-
rate, data = wavfile.read(audio)
|
43 |
-
|
44 |
-
# convert to mono
|
45 |
-
data = np.mean(data, axis=0)
|
46 |
-
|
47 |
-
# convert to float32
|
48 |
-
data = data.astype(np.float32)
|
49 |
-
|
50 |
-
# take a random 7 second slice of the audio
|
51 |
-
data = data[rate*7:rate*14]
|
52 |
-
|
53 |
-
spectrogram = spectrogram_from_waveform(
|
54 |
-
waveform=data,
|
55 |
-
sample_rate=rate,
|
56 |
-
# width=768,
|
57 |
-
n_fft=8192,
|
58 |
-
hop_length=512,
|
59 |
-
win_length=8192,
|
60 |
-
)
|
61 |
-
|
62 |
-
spec = image_from_spectrogram(spectrogram)
|
63 |
-
|
64 |
-
images = pipe2(
|
65 |
-
prompt=prompt,
|
66 |
-
image=spec,
|
67 |
-
strength=0.5,
|
68 |
-
guidance_scale=7
|
69 |
-
).images
|
70 |
-
|
71 |
-
wav = wav_bytes_from_spectrogram_image(images[0])
|
72 |
-
with open("output.wav", "wb") as f:
|
73 |
-
f.write(wav[0].getbuffer())
|
74 |
-
return images[0], 'output.wav', gr.update(visible=True), gr.update(visible=True), gr.update(visible=True)
|
75 |
|
76 |
title = """
|
77 |
<div style="text-align: center; max-width: 500px; margin: 0 auto;">
|
@@ -189,8 +142,10 @@ with gr.Blocks(css=css) as demo:
|
|
189 |
gr.HTML(title)
|
190 |
|
191 |
prompt_input = gr.Textbox(placeholder="a cat diva singing in a New York jazz club", label="Musical prompt", elem_id="prompt-in")
|
192 |
-
|
193 |
-
|
|
|
|
|
194 |
send_btn = gr.Button(value="Get a new spectrogram ! ", elem_id="submit-btn")
|
195 |
|
196 |
with gr.Column(elem_id="col-container-2"):
|
@@ -205,7 +160,7 @@ with gr.Blocks(css=css) as demo:
|
|
205 |
|
206 |
gr.HTML(article)
|
207 |
|
208 |
-
send_btn.click(predict, inputs=[prompt_input,
|
209 |
share_button.click(None, [], [], _js=share_js)
|
210 |
|
211 |
demo.queue(max_size=250).launch(debug=True)
|
|
|
1 |
import gradio as gr
|
2 |
import torch
|
3 |
|
4 |
+
from spectro import wav_bytes_from_spectrogram_image
|
|
|
|
|
|
|
|
|
5 |
|
6 |
from diffusers import StableDiffusionPipeline
|
7 |
from diffusers import StableDiffusionImg2ImgPipeline
|
|
|
11 |
MODEL_ID = "riffusion/riffusion-model-v1"
|
12 |
pipe = StableDiffusionPipeline.from_pretrained(MODEL_ID, torch_dtype=torch.float16)
|
13 |
pipe = pipe.to("cuda")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
14 |
|
15 |
+
|
16 |
+
def predict(prompt, negative_prompt, duration):
|
17 |
if duration == 5:
|
18 |
width_duration=512
|
19 |
else :
|
20 |
width_duration = 512 + ((int(duration)-5) * 128)
|
21 |
+
spec = pipe(prompt, negative_prompt=negative_prompt, height=512, width=width_duration).images[0]
|
22 |
print(spec)
|
23 |
wav = wav_bytes_from_spectrogram_image(spec)
|
24 |
with open("output.wav", "wb") as f:
|
25 |
f.write(wav[0].getbuffer())
|
26 |
return spec, 'output.wav', gr.update(visible=True), gr.update(visible=True), gr.update(visible=True)
|
27 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
28 |
|
29 |
title = """
|
30 |
<div style="text-align: center; max-width: 500px; margin: 0 auto;">
|
|
|
142 |
gr.HTML(title)
|
143 |
|
144 |
prompt_input = gr.Textbox(placeholder="a cat diva singing in a New York jazz club", label="Musical prompt", elem_id="prompt-in")
|
145 |
+
with gr.Row():
|
146 |
+
negative_prompt = gr.Textbox(label="Negative prompt")
|
147 |
+
duration_input = gr.Slider(label="Duration in seconds", minimum=5, maximum=10, step=1, value=8, elem_id="duration-slider")
|
148 |
+
|
149 |
send_btn = gr.Button(value="Get a new spectrogram ! ", elem_id="submit-btn")
|
150 |
|
151 |
with gr.Column(elem_id="col-container-2"):
|
|
|
160 |
|
161 |
gr.HTML(article)
|
162 |
|
163 |
+
send_btn.click(predict, inputs=[prompt_input, negative_prompt, duration_input], outputs=[spectrogram_output, sound_output, share_button, community_icon, loading_icon])
|
164 |
share_button.click(None, [], [], _js=share_js)
|
165 |
|
166 |
demo.queue(max_size=250).launch(debug=True)
|