Spaces:
Runtime error
Runtime error
Add application file
Browse files- app.py +127 -0
- asset/model/model_two.bin +3 -0
- asset/test/mix_clean/Audio0.wav +0 -0
- asset/test/mix_clean/Audio1.wav +0 -0
- asset/test/mix_clean/Audio2.wav +0 -0
- asset/test/mix_clean/Audio3.wav +0 -0
- asset/test/s1/Audio0.wav +0 -0
- asset/test/s1/Audio1.wav +0 -0
- asset/test/s1/Audio2.wav +0 -0
- asset/test/s1/Audio3.wav +0 -0
- asset/test/s2/Audio0.wav +0 -0
- asset/test/s2/Audio1.wav +0 -0
- asset/test/s2/Audio2.wav +0 -0
- asset/test/s2/Audio3.wav +0 -0
- requirements.txt +7 -0
app.py
ADDED
|
@@ -0,0 +1,127 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
# Importing libraries
|
| 3 |
+
|
| 4 |
+
import gradio as gr
|
| 5 |
+
from asteroid.models import ConvTasNet, DPRNNTasNet
|
| 6 |
+
import torch
|
| 7 |
+
import os
|
| 8 |
+
import shutil
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
# ------------------ #
|
| 12 |
+
|
| 13 |
+
class tester():
|
| 14 |
+
def __init__(self, model):
|
| 15 |
+
# Modeling
|
| 16 |
+
self.model = model
|
| 17 |
+
# Test Directory will also contain output after files
|
| 18 |
+
self.test_dir = '/asset/test_subject/'
|
| 19 |
+
def prepare_test(self):
|
| 20 |
+
# Removing older test and their results
|
| 21 |
+
if os.path.exists(self.test_dir):
|
| 22 |
+
shutil.rmtree(self.test_dir)
|
| 23 |
+
|
| 24 |
+
if not os.path.exists(self.test_dir):
|
| 25 |
+
os.mkdir(self.test_dir)
|
| 26 |
+
def test(self, path):
|
| 27 |
+
self.prepare_test()
|
| 28 |
+
|
| 29 |
+
test_subject = self.test_dir + 'test.wav'
|
| 30 |
+
shutil.copyfile(path, test_subject)
|
| 31 |
+
|
| 32 |
+
self.model.separate(test_subject,force_overwrite=True, resample=True)
|
| 33 |
+
|
| 34 |
+
def load_model():
|
| 35 |
+
model = torch.load('/asset/model/model_two.bin')
|
| 36 |
+
return model
|
| 37 |
+
|
| 38 |
+
def separator(original_audio, path):
|
| 39 |
+
Test = tester(load_model())
|
| 40 |
+
Test.prepare_test()
|
| 41 |
+
Test.test(path)
|
| 42 |
+
|
| 43 |
+
separated_audios = list()
|
| 44 |
+
separated_audios.append('/asset/test_subject/test.wav')
|
| 45 |
+
separated_audios.append('/asset/test_subject/test_est1.wav')
|
| 46 |
+
separated_audios.append('/asset/test_subject/test_est2.wav')
|
| 47 |
+
|
| 48 |
+
return separated_audios
|
| 49 |
+
|
| 50 |
+
demo = gr.Blocks(theme=gr.themes.Soft())
|
| 51 |
+
with demo:
|
| 52 |
+
|
| 53 |
+
gr.Markdown('''
|
| 54 |
+
<center>
|
| 55 |
+
<h1>Speech Separation</h1>
|
| 56 |
+
<div style="display:flex;align-items:center;justify-content:center;">
|
| 57 |
+
<iframe src="https://streamable.com/e/uribry?autoplay=1&nocontrols=1" frameborder="0" allow="autoplay">
|
| 58 |
+
</iframe>
|
| 59 |
+
</div>
|
| 60 |
+
<div></div>
|
| 61 |
+
<p>
|
| 62 |
+
It is a shareable demonstration window which can be used to view result on any device by setting 'share' a launch parameter 'True'.
|
| 63 |
+
It displays original audio for mixture of speaker, seperated audio by our model and original individual speaker audio.
|
| 64 |
+
</p>
|
| 65 |
+
</center>
|
| 66 |
+
''')
|
| 67 |
+
|
| 68 |
+
with gr.Row():
|
| 69 |
+
pass
|
| 70 |
+
with gr.Row():
|
| 71 |
+
pass
|
| 72 |
+
gr.Markdown('''
|
| 73 |
+
<h2> Original Audio</h2>
|
| 74 |
+
|
| 75 |
+
''')
|
| 76 |
+
with gr.Row():
|
| 77 |
+
output_text1 = gr.Text("Original Speech signal ", label='Original Audio', interactive=False)
|
| 78 |
+
original_audio = gr.Audio(label='Original Audio', interactive=False)
|
| 79 |
+
|
| 80 |
+
with gr.Row():
|
| 81 |
+
pass
|
| 82 |
+
with gr.Row():
|
| 83 |
+
pass
|
| 84 |
+
gr.Markdown('''
|
| 85 |
+
<h2> Separated Audio</h2>
|
| 86 |
+
|
| 87 |
+
''')
|
| 88 |
+
|
| 89 |
+
with gr.Row():
|
| 90 |
+
output_text1 = gr.Text("Separated Speech signal Speaker 1 ", label='Speaker 1', interactive=False)
|
| 91 |
+
output_audio1 = gr.Audio(label='Speaker 1', interactive=False)
|
| 92 |
+
with gr.Row():
|
| 93 |
+
output_text2 = gr.Text("Separated Speech signal Speaker 2 ", label='Speaker 2', interactive=False)
|
| 94 |
+
output_audio2 = gr.Audio(label='Speaker 2', interactive=False)
|
| 95 |
+
|
| 96 |
+
|
| 97 |
+
outputs_audio = [original_audio, output_audio1, output_audio2]
|
| 98 |
+
button = gr.Button("Separate")
|
| 99 |
+
examples = [
|
| 100 |
+
"/asset/test/mix_clean/Audio0.wav",
|
| 101 |
+
"/asset/test/mix_clean/Audio1.wav",
|
| 102 |
+
"/asset/test/mix_clean/Audio2.wav",
|
| 103 |
+
"/asset/test/mix_clean/Audio3.wav"
|
| 104 |
+
]
|
| 105 |
+
|
| 106 |
+
example_selector = gr.inputs.Radio(examples, label="Example Audio")
|
| 107 |
+
button.click(separator, inputs=[original_audio, example_selector], outputs=outputs_audio)
|
| 108 |
+
|
| 109 |
+
gr.Markdown('''
|
| 110 |
+
<center>
|
| 111 |
+
<div style="display:flex;align-items:center;justify-content:center;">
|
| 112 |
+
<a href="https://www.linkedin.com/in/dhruv73/" target="blank">
|
| 113 |
+
<img src="https://raw.githubusercontent.com/devicons/devicon/1119b9f84c0290e0f0b38982099a2bd027a48bf1/icons/linkedin/linkedin-original.svg" alt="LinkedIN: /dhruv_73" width="100" height="100"/>
|
| 114 |
+
</a>
|
| 115 |
+
 
|
| 116 |
+
<a href="https://github.com/DS-73" target="_blank">
|
| 117 |
+
<img src="https://raw.githubusercontent.com/devicons/devicon/1119b9f84c0290e0f0b38982099a2bd027a48bf1/icons/github/github-original.svg" alt="Github: /DS-73" width="100" height="100"/>
|
| 118 |
+
</a>
|
| 119 |
+
</div>
|
| 120 |
+
</center>
|
| 121 |
+
''')
|
| 122 |
+
|
| 123 |
+
demo.launch()
|
| 124 |
+
|
| 125 |
+
|
| 126 |
+
|
| 127 |
+
# ------------------ #
|
asset/model/model_two.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1a6b78dc3ba01e592970d9ce9cda9a1ce0ac003d000aad93ae6833e254a36612
|
| 3 |
+
size 20410329
|
asset/test/mix_clean/Audio0.wav
ADDED
|
Binary file (88.4 kB). View file
|
|
|
asset/test/mix_clean/Audio1.wav
ADDED
|
Binary file (139 kB). View file
|
|
|
asset/test/mix_clean/Audio2.wav
ADDED
|
Binary file (225 kB). View file
|
|
|
asset/test/mix_clean/Audio3.wav
ADDED
|
Binary file (194 kB). View file
|
|
|
asset/test/s1/Audio0.wav
ADDED
|
Binary file (88.4 kB). View file
|
|
|
asset/test/s1/Audio1.wav
ADDED
|
Binary file (139 kB). View file
|
|
|
asset/test/s1/Audio2.wav
ADDED
|
Binary file (225 kB). View file
|
|
|
asset/test/s1/Audio3.wav
ADDED
|
Binary file (194 kB). View file
|
|
|
asset/test/s2/Audio0.wav
ADDED
|
Binary file (88.4 kB). View file
|
|
|
asset/test/s2/Audio1.wav
ADDED
|
Binary file (139 kB). View file
|
|
|
asset/test/s2/Audio2.wav
ADDED
|
Binary file (225 kB). View file
|
|
|
asset/test/s2/Audio3.wav
ADDED
|
Binary file (194 kB). View file
|
|
|
requirements.txt
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
gradio==3.34.0
|
| 2 |
+
gradio_client==0.2.6
|
| 3 |
+
torch==2.0.1
|
| 4 |
+
asteroid==0.6.0
|
| 5 |
+
asteroid-filterbanks==0.4.0
|
| 6 |
+
os
|
| 7 |
+
shutil
|