Spaces:
Build error
Build error
razaraees10
commited on
Commit
•
2355e92
1
Parent(s):
f9a3150
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,165 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from TTS.api import TTS
|
2 |
+
tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2", gpu=True).to("cuda")
|
3 |
+
# Dependencies
|
4 |
+
%cd /content/
|
5 |
+
|
6 |
+
import locale
|
7 |
+
locale.getpreferredencoding = lambda: "UTF-8"
|
8 |
+
|
9 |
+
!git clone https://github.com/justinjohn0306/Wav2Lip
|
10 |
+
!cd Wav2Lip && pip install -r requirements_colab.txt
|
11 |
+
|
12 |
+
%cd /content/Wav2Lip
|
13 |
+
|
14 |
+
!wget "https://www.adrianbulat.com/downloads/python-fan/s3fd-619a316812.pth" -O "face_detection/detection/sfd/s3fd.pth"
|
15 |
+
!wget 'https://github.com/justinjohn0306/Wav2Lip/releases/download/models/wav2lip.pth' -O 'checkpoints/wav2lip.pth'
|
16 |
+
!wget 'https://github.com/justinjohn0306/Wav2Lip/releases/download/models/wav2lip_gan.pth' -O 'checkpoints/wav2lip_gan.pth'
|
17 |
+
!wget 'https://github.com/justinjohn0306/Wav2Lip/releases/download/models/resnet50.pth' -O 'checkpoints/resnet50.pth'
|
18 |
+
!wget 'https://github.com/justinjohn0306/Wav2Lip/releases/download/models/mobilenet.pth' -O 'checkpoints/mobilenet.pth'
|
19 |
+
import subprocess
|
20 |
+
import assemblyai as aai
|
21 |
+
import requests
|
22 |
+
import moviepy.editor as mp
|
23 |
+
|
24 |
+
class translation:
|
25 |
+
def __init__(self,video_path,original_language,target_language):
|
26 |
+
self.video_path= video_path
|
27 |
+
self.original_language = original_language
|
28 |
+
self.target_language = target_language
|
29 |
+
|
30 |
+
|
31 |
+
|
32 |
+
def org_language_parameters(self,original_language):
|
33 |
+
if original_language == 'English':
|
34 |
+
self.lan_code='en'
|
35 |
+
elif original_language =='German':
|
36 |
+
self.lan_code='de'
|
37 |
+
elif original_language =='French':
|
38 |
+
self.lan_code='fr'
|
39 |
+
elif original_language =='Spanish':
|
40 |
+
self.lan_code='es'
|
41 |
+
else:
|
42 |
+
self.lan_code = ''
|
43 |
+
|
44 |
+
def target_language_parameters(self,target_language):
|
45 |
+
if target_language == 'English':
|
46 |
+
self.tran_code='en'
|
47 |
+
elif target_language =='German':
|
48 |
+
self.tran_code='de'
|
49 |
+
elif target_language =='French':
|
50 |
+
self.tran_code='fr'
|
51 |
+
elif target_language =='Spanish':
|
52 |
+
self.tran_code='es'
|
53 |
+
else:
|
54 |
+
self.tran_code = ''
|
55 |
+
|
56 |
+
def extract_audio(self):
|
57 |
+
video = mp.VideoFileClip(self.video_path)
|
58 |
+
audio = video.audio
|
59 |
+
audio_path = "output_audio.wav"
|
60 |
+
audio.write_audiofile(audio_path)
|
61 |
+
print("Audio extracted successfully!")
|
62 |
+
return audio_path
|
63 |
+
|
64 |
+
|
65 |
+
def transcribe_audio(self,audio_path):
|
66 |
+
aai.settings.api_key = "c29eb650444a4ae4be6a787ebb15d5e2"
|
67 |
+
config = aai.TranscriptionConfig(language_code=self.lan_code)
|
68 |
+
transcriber = aai.Transcriber(config=config)
|
69 |
+
transcript = transcriber.transcribe(audio_path)
|
70 |
+
transcript_text = transcript.text
|
71 |
+
return transcript_text
|
72 |
+
if transcript.status == aai.TranscriptStatus.error:
|
73 |
+
print(transcript.error)
|
74 |
+
return None
|
75 |
+
|
76 |
+
|
77 |
+
def translate_text(self,transcript_text):
|
78 |
+
base_url = "https://api.cognitive.microsofttranslator.com"
|
79 |
+
endpoint = "/translate"
|
80 |
+
headers = {
|
81 |
+
"Ocp-Apim-Subscription-Key": "cd226bb1f3644276bea01d82dd861cbb",
|
82 |
+
"Content-Type": "application/json",
|
83 |
+
"Ocp-Apim-Subscription-Region": "southeastasia"
|
84 |
+
}
|
85 |
+
params = {
|
86 |
+
"api-version": "3.0",
|
87 |
+
"from": self.lan_code,
|
88 |
+
"to": self.tran_code
|
89 |
+
}
|
90 |
+
body = [{"text": transcript_text}]
|
91 |
+
response = requests.post(base_url + endpoint, headers=headers, params=params, json=body)
|
92 |
+
response.raise_for_status()
|
93 |
+
translation = response.json()[0]["translations"][0]["text"]
|
94 |
+
return translation
|
95 |
+
|
96 |
+
|
97 |
+
|
98 |
+
|
99 |
+
#generate audio
|
100 |
+
def generate_audio(self,translated_text):
|
101 |
+
tts.tts_to_file(translated_text,
|
102 |
+
speaker_wav='output_audio.wav',
|
103 |
+
file_path="output_synth.wav",
|
104 |
+
language= self.tran_code
|
105 |
+
)
|
106 |
+
return "output_synth.wav"
|
107 |
+
|
108 |
+
def translate_video(self):
|
109 |
+
audio_path = self.extract_audio()
|
110 |
+
self.org_language_parameters(self.original_language)
|
111 |
+
self.target_language_parameters(self.target_language)
|
112 |
+
transcript_text = self.transcribe_audio(audio_path)
|
113 |
+
translated_text = self.translate_text(transcript_text)
|
114 |
+
translated_audio_path = self.generate_audio(translated_text)
|
115 |
+
#Generate video
|
116 |
+
%cd /content/Wav2Lip
|
117 |
+
|
118 |
+
#This is the detection box padding, if you see it doesnt sit quite right, just adjust the values a bit. Usually the bottom one is the biggest issue
|
119 |
+
pad_top = 0
|
120 |
+
pad_bottom = 15
|
121 |
+
pad_left = 0
|
122 |
+
pad_right = 0
|
123 |
+
rescaleFactor = 1
|
124 |
+
video_path_fix = f"'../{self.video_path}'"
|
125 |
+
audio_path_fix = f"'../{translated_audio_path}'"
|
126 |
+
!python inference.py --checkpoint_path 'checkpoints/wav2lip_gan.pth' --face $video_path_fix --audio $audio_path_fix --pads $pad_top $pad_bottom $pad_left $pad_right --resize_factor $rescaleFactor --nosmooth --outfile '/content/output_video.mp4'
|
127 |
+
return '/content/output_video.mp4' # Return the path to the translated video file
|
128 |
+
# return '/content/output_video.mp4', open('/content/output_video.mp4', 'rb') # Return the path and file object of the translated video file
|
129 |
+
|
130 |
+
# from translator import translation # Import the Translator class from translate module
|
131 |
+
import gradio as gr
|
132 |
+
import os
|
133 |
+
from google.colab import files
|
134 |
+
|
135 |
+
|
136 |
+
def app(video_path, original_language, target_language):
|
137 |
+
%cd /content/
|
138 |
+
|
139 |
+
video_name = os.path.basename(video_path)
|
140 |
+
|
141 |
+
# Save the uploaded file to the content folder in Colab
|
142 |
+
with open(video_name, "wb") as f:
|
143 |
+
with open(video_path, "rb") as uploaded_file:
|
144 |
+
f.write(uploaded_file.read())
|
145 |
+
|
146 |
+
|
147 |
+
translator = translation(video_name, original_language, target_language)
|
148 |
+
video_file = translator.translate_video()
|
149 |
+
|
150 |
+
|
151 |
+
return video_file
|
152 |
+
|
153 |
+
interface_video_file = gr.Interface(
|
154 |
+
fn=app,
|
155 |
+
inputs=[
|
156 |
+
gr.Video(label="Video Path"),
|
157 |
+
gr.Dropdown(["English", "German", "French", "Spanish"], label="Original Language"),
|
158 |
+
gr.Dropdown(["English", "German", "French", "Spanish"], label="Targeted Language"),
|
159 |
+
],
|
160 |
+
outputs=gr.Video(label="Translated Video")
|
161 |
+
)
|
162 |
+
|
163 |
+
interface_video_file.launch(debug=True)
|
164 |
+
|
165 |
+
|