Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -1,36 +1,17 @@
|
|
1 |
import gradio as gr
|
2 |
-
#import torch
|
3 |
-
#import whisper
|
4 |
-
#from datetime import datetime
|
5 |
from PIL import Image
|
6 |
-
#import flag
|
7 |
import os
|
8 |
-
#MY_SECRET_TOKEN=os.environ.get('HF_TOKEN_SD')
|
9 |
|
10 |
#from diffusers import StableDiffusionPipeline
|
11 |
-
|
12 |
whisper = gr.Interface.load(name="spaces/sanchit-gandhi/whisper-large-v2")
|
13 |
stable_diffusion = gr.Blocks.load(name="spaces/runwayml/stable-diffusion-v1-5")
|
14 |
### ββββββββββββββββββββββββββββββββββββββββ
|
15 |
-
|
16 |
-
title="Whisper to Stable Diffusion"
|
17 |
-
|
18 |
### ββββββββββββββββββββββββββββββββββββββββ
|
19 |
-
|
20 |
-
#whisper_model = whisper.load_model("small")
|
21 |
-
|
22 |
-
#device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
|
23 |
-
|
24 |
-
#pipe = StableDiffusionPipeline.from_pretrained("CompVis/stable-diffusion-v1-4", use_auth_token=MY_SECRET_TOKEN)
|
25 |
-
#pipe.to(device)
|
26 |
-
|
27 |
-
### ββββββββββββββββββββββββββββββββββββββββ
|
28 |
-
|
29 |
def get_images(prompt):
|
30 |
gallery_dir = stable_diffusion(prompt, fn_index=2)
|
31 |
return [os.path.join(gallery_dir, img) for img in os.listdir(gallery_dir)]
|
32 |
|
33 |
-
|
34 |
def magic_whisper_to_sd(audio, guidance_scale, nb_iterations, seed):
|
35 |
|
36 |
whisper_results = translate_better(audio)
|
@@ -39,42 +20,7 @@ def magic_whisper_to_sd(audio, guidance_scale, nb_iterations, seed):
|
|
39 |
|
40 |
return whisper_results[0], whisper_results[1], images
|
41 |
|
42 |
-
|
43 |
-
#
|
44 |
-
# generator = torch.Generator(device=device).manual_seed(int(seed))
|
45 |
-
#
|
46 |
-
# print("""
|
47 |
-
# β
|
48 |
-
# Sending prompt to Stable Diffusion ...
|
49 |
-
# β
|
50 |
-
# """)
|
51 |
-
# print("prompt: " + prompt)
|
52 |
-
# print("guidance scale: " + str(guidance_scale))
|
53 |
-
# print("inference steps: " + str(nb_iterations))
|
54 |
-
# print("seed: " + str(seed))
|
55 |
-
#
|
56 |
-
# images_list = pipe(
|
57 |
-
# [prompt] * 2,
|
58 |
-
# guidance_scale=guidance_scale,
|
59 |
-
# num_inference_steps=nb_iterations,
|
60 |
-
# generator=generator
|
61 |
-
# )
|
62 |
-
#
|
63 |
-
# images = []
|
64 |
-
#
|
65 |
-
# safe_image = Image.open(r"unsafe.png")
|
66 |
-
#
|
67 |
-
# for i, image in enumerate(images_list["sample"]):
|
68 |
-
# if(images_list["nsfw_content_detected"][i]):
|
69 |
-
# images.append(safe_image)
|
70 |
-
# else:
|
71 |
-
# images.append(image)
|
72 |
-
#
|
73 |
-
#
|
74 |
-
# print("Stable Diffusion has finished")
|
75 |
-
# print("βββββββββββββββββββββββββββββββββββββββββββ")
|
76 |
-
#
|
77 |
-
# return images
|
78 |
|
79 |
def translate_better(audio):
|
80 |
print("""
|
@@ -91,227 +37,14 @@ def translate_better(audio):
|
|
91 |
return transcribe_text_result, translate_text_result
|
92 |
|
93 |
|
94 |
-
#def translate(audio):
|
95 |
-
# print("""
|
96 |
-
# β
|
97 |
-
# Sending audio to Whisper ...
|
98 |
-
# β
|
99 |
-
# """)
|
100 |
-
# # current dateTime
|
101 |
-
# now = datetime.now()
|
102 |
-
# # convert to string
|
103 |
-
# date_time_str = now.strftime("%Y-%m-%d %H:%M:%S")
|
104 |
-
# print('DateTime String:', date_time_str)
|
105 |
-
#
|
106 |
-
# audio = whisper.load_audio(audio)
|
107 |
-
# audio = whisper.pad_or_trim(audio)
|
108 |
-
#
|
109 |
-
# mel = whisper.log_mel_spectrogram(audio).to(whisper_model.device)
|
110 |
-
#
|
111 |
-
# _, probs = whisper_model.detect_language(mel)
|
112 |
-
#
|
113 |
-
# transcript_options = whisper.DecodingOptions(task="transcribe", fp16 = False)
|
114 |
-
# translate_options = whisper.DecodingOptions(task="translate", fp16 = False)
|
115 |
-
#
|
116 |
-
# transcription = whisper.decode(whisper_model, mel, transcript_options)
|
117 |
-
# translation = whisper.decode(whisper_model, mel, translate_options)
|
118 |
-
#
|
119 |
-
# print("language spoken: " + transcription.language)
|
120 |
-
# print("transcript: " + transcription.text)
|
121 |
-
# print("βββββββββββββββββββββββββββββββββββββββββββ")
|
122 |
-
# print("translated: " + translation.text)
|
123 |
-
# if transcription.language == "en":
|
124 |
-
# tr_flag = flag.flag('GB')
|
125 |
-
# else:
|
126 |
-
# tr_flag = flag.flag(transcription.language)
|
127 |
-
# return tr_flag, transcription.text, translation.text
|
128 |
-
|
129 |
-
### ββββββββββββββββββββββββββββββββββββββββ
|
130 |
-
|
131 |
-
css = """
|
132 |
-
.container {
|
133 |
-
max-width: 780px;
|
134 |
-
margin: auto;
|
135 |
-
padding-top: 1.5rem;
|
136 |
-
}
|
137 |
-
a {
|
138 |
-
text-decoration: underline;
|
139 |
-
}
|
140 |
-
h1 {
|
141 |
-
font-weight: 900;
|
142 |
-
margin-bottom: 7px;
|
143 |
-
text-align: center;
|
144 |
-
font-size: 2em;
|
145 |
-
margin-bottom: 1em;
|
146 |
-
}
|
147 |
-
#w2sd_container{
|
148 |
-
margin-top: 20px;
|
149 |
-
}
|
150 |
-
.footer {
|
151 |
-
margin-bottom: 45px;
|
152 |
-
margin-top: 35px;
|
153 |
-
text-align: center;
|
154 |
-
border-bottom: 1px solid #e5e5e5;
|
155 |
-
}
|
156 |
-
.footer>p {
|
157 |
-
font-size: .8rem;
|
158 |
-
display: inline-block;
|
159 |
-
padding: 0 10px;
|
160 |
-
transform: translateY(10px);
|
161 |
-
background: white;
|
162 |
-
}
|
163 |
-
.dark .footer {
|
164 |
-
border-color: #303030;
|
165 |
-
}
|
166 |
-
.dark .footer>p {
|
167 |
-
background: #0b0f19;
|
168 |
-
}
|
169 |
-
.tabitem {
|
170 |
-
border-bottom-left-radius: 10px;
|
171 |
-
border-bottom-right-radius: 10px;
|
172 |
-
}
|
173 |
-
#record_tab, #upload_tab {
|
174 |
-
font-size: 1.2em;
|
175 |
-
}
|
176 |
-
#record_btn{
|
177 |
-
|
178 |
-
}
|
179 |
-
#record_btn > div > button > span {
|
180 |
-
width: 2.375rem;
|
181 |
-
height: 2.375rem;
|
182 |
-
}
|
183 |
-
#record_btn > div > button > span > span {
|
184 |
-
width: 2.375rem;
|
185 |
-
height: 2.375rem;
|
186 |
-
}
|
187 |
-
audio {
|
188 |
-
margin-bottom: 10px;
|
189 |
-
}
|
190 |
-
div#record_btn > .mt-6{
|
191 |
-
margin-top: 0!important;
|
192 |
-
}
|
193 |
-
div#record_btn > .mt-6 button {
|
194 |
-
font-size: 2em;
|
195 |
-
width: 100%;
|
196 |
-
padding: 20px;
|
197 |
-
height: 160px;
|
198 |
-
}
|
199 |
-
div#upload_area {
|
200 |
-
height: 11.1rem;
|
201 |
-
}
|
202 |
-
div#upload_area > div.w-full > div {
|
203 |
-
min-height: 9rem;
|
204 |
-
}
|
205 |
-
#check_btn_1, #check_btn_2{
|
206 |
-
color: #fff;
|
207 |
-
--tw-gradient-from: #4caf50;
|
208 |
-
--tw-gradient-stops: var(--tw-gradient-from), var(--tw-gradient-to);
|
209 |
-
--tw-gradient-to: #4caf50;
|
210 |
-
border-color: #8bc34a;
|
211 |
-
}
|
212 |
-
#magic_btn_1, #magic_btn_2{
|
213 |
-
color: #fff;
|
214 |
-
--tw-gradient-from: #f44336;
|
215 |
-
--tw-gradient-stops: var(--tw-gradient-from), var(--tw-gradient-to);
|
216 |
-
--tw-gradient-to: #ff9800;
|
217 |
-
border-color: #ff9800;
|
218 |
-
}
|
219 |
-
input::-webkit-inner-spin-button, input::-webkit-outer-spin-button {
|
220 |
-
-webkit-appearance: none;
|
221 |
-
}
|
222 |
-
input[type=number] {
|
223 |
-
-moz-appearance: textfield;
|
224 |
-
}
|
225 |
-
input[type=range] {
|
226 |
-
-webkit-appearance: none;
|
227 |
-
cursor: pointer;
|
228 |
-
height: 1px;
|
229 |
-
background: currentColor;
|
230 |
-
}
|
231 |
-
input[type=range]::-webkit-slider-thumb {
|
232 |
-
-webkit-appearance: none;
|
233 |
-
width: 0.5em;
|
234 |
-
height: 1.2em;
|
235 |
-
border-radius: 10px;
|
236 |
-
background: currentColor;
|
237 |
-
}
|
238 |
-
input[type=range]::-moz-range-thumb{
|
239 |
-
width: 0.5em;
|
240 |
-
height: 1.2em;
|
241 |
-
border-radius: 10px;
|
242 |
-
background: currentColor;
|
243 |
-
}
|
244 |
-
div#spoken_lang textarea {
|
245 |
-
font-size: 4em;
|
246 |
-
line-height: 1em;
|
247 |
-
text-align: center;
|
248 |
-
}
|
249 |
-
div#transcripted {
|
250 |
-
flex: 4;
|
251 |
-
}
|
252 |
-
div#translated textarea {
|
253 |
-
font-size: 1.5em;
|
254 |
-
line-height: 1.25em;
|
255 |
-
}
|
256 |
-
#sd_settings {
|
257 |
-
margin-bottom: 20px;
|
258 |
-
}
|
259 |
-
#diffuse_btn {
|
260 |
-
color: #fff;
|
261 |
-
font-size: 1em;
|
262 |
-
margin-bottom: 20px;
|
263 |
-
--tw-gradient-from: #4caf50;
|
264 |
-
--tw-gradient-stops: var(--tw-gradient-from), var(--tw-gradient-to);
|
265 |
-
--tw-gradient-to: #4caf50;
|
266 |
-
border-color: #8bc34a;
|
267 |
-
}
|
268 |
-
#notice {
|
269 |
-
padding: 20px 14px 10px;
|
270 |
-
display: flex;
|
271 |
-
align-content: space-evenly;
|
272 |
-
gap: 20px;
|
273 |
-
line-height: 1em;
|
274 |
-
font-size: .8em;
|
275 |
-
border: 1px solid #374151;
|
276 |
-
border-radius: 10px;
|
277 |
-
}
|
278 |
-
#about {
|
279 |
-
padding: 20px;
|
280 |
-
}
|
281 |
-
#notice > div {
|
282 |
-
flex: 1;
|
283 |
-
}
|
284 |
-
|
285 |
-
"""
|
286 |
-
|
287 |
-
### ββββββββββββββββββββββββββββββββββββββββ
|
288 |
|
289 |
with gr.Blocks(css=css) as demo:
|
290 |
-
|
291 |
-
gr.HTML('''
|
292 |
-
<h1>
|
293 |
-
Whisper to Stable Diffusion
|
294 |
-
</h1>
|
295 |
-
<p style='text-align: center;'>
|
296 |
-
Ask stable diffusion for images by speaking (or singing π€) in your native language ! Try it in French π
|
297 |
-
</p>
|
298 |
-
|
299 |
-
<p style='text-align: center;'>
|
300 |
-
This demo is wired to the official SD Space β’ Offered by Sylvain <a href='https://twitter.com/fffiloni' target='_blank'>@fffiloni</a> β’ <img id='visitor-badge' alt='visitor badge' src='https://visitor-badge.glitch.me/badge?page_id=gradio-blocks.whisper-to-stable-diffusion' style='display: inline-block' /><br />
|
301 |
-
β
|
302 |
-
</p>
|
303 |
-
|
304 |
-
''')
|
305 |
-
# with gr.Row(elem_id="w2sd_container"):
|
306 |
-
# with gr.Column():
|
307 |
-
|
308 |
-
gr.Markdown(
|
309 |
"""
|
310 |
-
|
311 |
-
## 1. Record audio or Upload an audio file:
|
312 |
"""
|
313 |
)
|
314 |
-
|
315 |
with gr.Tab(label="Record audio input", elem_id="record_tab"):
|
316 |
with gr.Column():
|
317 |
record_input = gr.Audio(
|
@@ -369,13 +102,7 @@ with gr.Blocks(css=css) as demo:
|
|
369 |
|
370 |
clear_btn.click(fn=lambda value: gr.update(value=""), inputs=clear_btn, outputs=translated_output)
|
371 |
|
372 |
-
|
373 |
-
|
374 |
-
|
375 |
-
|
376 |
-
# with gr.Column():
|
377 |
-
|
378 |
-
|
379 |
|
380 |
gr.Markdown("""
|
381 |
## 3. Wait for Stable Diffusion Results βοΈ
|
@@ -386,42 +113,6 @@ with gr.Blocks(css=css) as demo:
|
|
386 |
sd_output = gr.Gallery().style(grid=2, height="auto")
|
387 |
|
388 |
|
389 |
-
gr.Markdown("""
|
390 |
-
### π About the models
|
391 |
-
<p style='font-size: 1em;line-height: 1.5em;'>
|
392 |
-
<strong>Whisper</strong> is a general-purpose speech recognition model.<br /><br />
|
393 |
-
It is trained on a large dataset of diverse audio and is also a multi-task model that can perform multilingual speech recognition as well as speech translation and language identification. <br />
|
394 |
-
β
|
395 |
-
</p>
|
396 |
-
<p style='font-size: 1em;line-height: 1.5em;'>
|
397 |
-
<strong>Stable Diffusion</strong> is a state of the art text-to-image model that generates images from text.
|
398 |
-
</p>
|
399 |
-
<div id="notice">
|
400 |
-
<div>
|
401 |
-
LICENSE
|
402 |
-
<p style='font-size: 0.8em;'>
|
403 |
-
The model is licensed with a <a href="https://huggingface.co/spaces/CompVis/stable-diffusion-license" target="_blank">CreativeML Open RAIL-M</a> license.</p>
|
404 |
-
<p style='font-size: 0.8em;'>
|
405 |
-
The authors claim no rights on the outputs you generate, you are free to use them and are accountable for their use which must not go against the provisions set in this license.</p>
|
406 |
-
<p style='font-size: 0.8em;'>
|
407 |
-
The license forbids you from sharing any content that violates any laws, produce any harm to a person, disseminate any personal information that would be meant for harm, spread misinformation and target vulnerable groups.</p>
|
408 |
-
<p style='font-size: 0.8em;'>
|
409 |
-
For the full list of restrictions please <a href="https://huggingface.co/spaces/CompVis/stable-diffusion-license" target="_blank" target="_blank">read the license</a>.
|
410 |
-
</p>
|
411 |
-
</div>
|
412 |
-
<div>
|
413 |
-
Biases and content acknowledgment
|
414 |
-
<p style='font-size: 0.8em;'>
|
415 |
-
Despite how impressive being able to turn text into image is, beware to the fact that this model may output content that reinforces or exacerbates societal biases, as well as realistic faces, pornography and violence.</p>
|
416 |
-
<p style='font-size: 0.8em;'>
|
417 |
-
The model was trained on the <a href="https://laion.ai/blog/laion-5b/" target="_blank">LAION-5B dataset</a>, which scraped non-curated image-text-pairs from the internet (the exception being the removal of illegal content) and is meant for research purposes.</p>
|
418 |
-
<p style='font-size: 0.8em;'> You can read more in the <a href="https://huggingface.co/CompVis/stable-diffusion-v1-4" target="_blank">model card</a>.
|
419 |
-
</p>
|
420 |
-
</div>
|
421 |
-
</div>
|
422 |
-
|
423 |
-
""", elem_id="about")
|
424 |
-
|
425 |
audio_r_translate.click(translate_better,
|
426 |
inputs = record_input,
|
427 |
outputs = [
|
@@ -472,13 +163,6 @@ with gr.Blocks(css=css) as demo:
|
|
472 |
],
|
473 |
outputs = sd_output
|
474 |
)
|
475 |
-
gr.HTML('''
|
476 |
-
<div class="footer">
|
477 |
-
<p>Whisper by <a href="https://github.com/openai/whisper" target="_blank">OpenAI</a> - Stable Diffusion by <a href="https://huggingface.co/CompVis" target="_blank">CompVis</a> and <a href="https://huggingface.co/stabilityai" target="_blank">Stability AI</a>
|
478 |
-
</p>
|
479 |
-
</div>
|
480 |
-
''')
|
481 |
-
|
482 |
|
483 |
if __name__ == "__main__":
|
484 |
demo.queue(max_size=32, concurrency_count=20).launch()
|
|
|
1 |
import gradio as gr
|
|
|
|
|
|
|
2 |
from PIL import Image
|
|
|
3 |
import os
|
|
|
4 |
|
5 |
#from diffusers import StableDiffusionPipeline
|
|
|
6 |
whisper = gr.Interface.load(name="spaces/sanchit-gandhi/whisper-large-v2")
|
7 |
stable_diffusion = gr.Blocks.load(name="spaces/runwayml/stable-diffusion-v1-5")
|
8 |
### ββββββββββββββββββββββββββββββββββββββββ
|
9 |
+
title="Talking to Stable Diffusion"
|
|
|
|
|
10 |
### ββββββββββββββββββββββββββββββββββββββββ
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
def get_images(prompt):
|
12 |
gallery_dir = stable_diffusion(prompt, fn_index=2)
|
13 |
return [os.path.join(gallery_dir, img) for img in os.listdir(gallery_dir)]
|
14 |
|
|
|
15 |
def magic_whisper_to_sd(audio, guidance_scale, nb_iterations, seed):
|
16 |
|
17 |
whisper_results = translate_better(audio)
|
|
|
20 |
|
21 |
return whisper_results[0], whisper_results[1], images
|
22 |
|
23 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
24 |
|
25 |
def translate_better(audio):
|
26 |
print("""
|
|
|
37 |
return transcribe_text_result, translate_text_result
|
38 |
|
39 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
40 |
|
41 |
with gr.Blocks(css=css) as demo:
|
42 |
+
gr.Markdown(
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
43 |
"""
|
44 |
+
## 1. Say what you want:
|
|
|
45 |
"""
|
46 |
)
|
47 |
+
with gr.Column():
|
48 |
with gr.Tab(label="Record audio input", elem_id="record_tab"):
|
49 |
with gr.Column():
|
50 |
record_input = gr.Audio(
|
|
|
102 |
|
103 |
clear_btn.click(fn=lambda value: gr.update(value=""), inputs=clear_btn, outputs=translated_output)
|
104 |
|
105 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
106 |
|
107 |
gr.Markdown("""
|
108 |
## 3. Wait for Stable Diffusion Results βοΈ
|
|
|
113 |
sd_output = gr.Gallery().style(grid=2, height="auto")
|
114 |
|
115 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
116 |
audio_r_translate.click(translate_better,
|
117 |
inputs = record_input,
|
118 |
outputs = [
|
|
|
163 |
],
|
164 |
outputs = sd_output
|
165 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
166 |
|
167 |
if __name__ == "__main__":
|
168 |
demo.queue(max_size=32, concurrency_count=20).launch()
|