Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
|
@@ -2,7 +2,8 @@ import os
|
|
| 2 |
import requests
|
| 3 |
import gradio as gr
|
| 4 |
import moviepy.editor as mp
|
| 5 |
-
from TTS.
|
|
|
|
| 6 |
import torch
|
| 7 |
import assemblyai as aai
|
| 8 |
|
|
@@ -15,27 +16,7 @@ model_files = {
|
|
| 15 |
"s3fd.pth": "https://www.adrianbulat.com/downloads/python-fan/s3fd-619a316812.pth"
|
| 16 |
}
|
| 17 |
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
# Get the path to the xtts_v2 folder
|
| 21 |
-
tts = os.path.join(os.getcwd(), 'xtts_v2')
|
| 22 |
-
|
| 23 |
-
# List all files in xtts_v2
|
| 24 |
-
files = os.listdir(tts)
|
| 25 |
-
print("Files in xtts_v2:", files)
|
| 26 |
-
|
| 27 |
-
# Iterate through the files in xtts_v2
|
| 28 |
-
for file_name in files:
|
| 29 |
-
file_path = os.path.join(tts, file_name)
|
| 30 |
-
|
| 31 |
-
# Check if it's a file or directory
|
| 32 |
-
if os.path.isfile(file_path):
|
| 33 |
-
print(f"{file_name} is a file.")
|
| 34 |
-
elif os.path.isdir(file_path):
|
| 35 |
-
print(f"{file_name} is a directory.")
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
for filename, url in model_files.items():
|
| 40 |
file_path = os.path.join("checkpoints" if "pth" in filename else "face_detection", filename)
|
| 41 |
if not os.path.exists(file_path):
|
|
@@ -44,15 +25,35 @@ for filename, url in model_files.items():
|
|
| 44 |
with open(file_path, 'wb') as f:
|
| 45 |
f.write(r.content)
|
| 46 |
|
| 47 |
-
# Initialize
|
| 48 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 49 |
|
| 50 |
# Translation class
|
| 51 |
-
class
|
| 52 |
def __init__(self, video_path, original_language, target_language):
|
| 53 |
self.video_path = video_path
|
| 54 |
self.original_language = original_language
|
| 55 |
self.target_language = target_language
|
|
|
|
| 56 |
|
| 57 |
def org_language_parameters(self, original_language):
|
| 58 |
language_codes = {'English': 'en', 'German': 'de', 'Italian': 'it', 'Spanish': 'es'}
|
|
@@ -90,8 +91,25 @@ class translation:
|
|
| 90 |
return translation
|
| 91 |
|
| 92 |
def generate_audio(self, translated_text):
|
| 93 |
-
|
| 94 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 95 |
|
| 96 |
def translate_video(self):
|
| 97 |
audio_path = self.extract_audio()
|
|
@@ -105,10 +123,9 @@ class translation:
|
|
| 105 |
os.system(f"python inference.py --checkpoint_path 'checkpoints/wav2lip_gan.pth' --face {self.video_path} --audio {translated_audio_path} --outfile 'output_video.mp4'")
|
| 106 |
return 'output_video.mp4'
|
| 107 |
|
| 108 |
-
|
| 109 |
# Gradio Interface
|
| 110 |
def app(video_path, original_language, target_language):
|
| 111 |
-
translator =
|
| 112 |
video_file = translator.translate_video()
|
| 113 |
return video_file
|
| 114 |
|
|
@@ -122,4 +139,4 @@ interface = gr.Interface(
|
|
| 122 |
outputs=gr.Video(label="Translated Video")
|
| 123 |
)
|
| 124 |
|
| 125 |
-
interface.launch()
|
|
|
|
| 2 |
import requests
|
| 3 |
import gradio as gr
|
| 4 |
import moviepy.editor as mp
|
| 5 |
+
from TTS.tts.configs.xtts_config import XttsConfig
|
| 6 |
+
from TTS.tts.models.xtts import Xtts
|
| 7 |
import torch
|
| 8 |
import assemblyai as aai
|
| 9 |
|
|
|
|
| 16 |
"s3fd.pth": "https://www.adrianbulat.com/downloads/python-fan/s3fd-619a316812.pth"
|
| 17 |
}
|
| 18 |
|
| 19 |
+
# Download model files
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
for filename, url in model_files.items():
|
| 21 |
file_path = os.path.join("checkpoints" if "pth" in filename else "face_detection", filename)
|
| 22 |
if not os.path.exists(file_path):
|
|
|
|
| 25 |
with open(file_path, 'wb') as f:
|
| 26 |
f.write(r.content)
|
| 27 |
|
| 28 |
+
# Initialize xtts model
|
| 29 |
+
def initialize_xtts_model():
|
| 30 |
+
# Get the path to the xtts_v2 folder
|
| 31 |
+
tts_dir = os.path.join(os.getcwd(), 'xtts_v2')
|
| 32 |
+
|
| 33 |
+
# Load the configuration
|
| 34 |
+
config_path = os.path.join(tts_dir, 'config.json')
|
| 35 |
+
config = XttsConfig()
|
| 36 |
+
config.load_json(config_path)
|
| 37 |
+
|
| 38 |
+
# Initialize the model from the configuration
|
| 39 |
+
model = Xtts.init_from_config(config)
|
| 40 |
+
|
| 41 |
+
# Load the model checkpoint
|
| 42 |
+
model.load_checkpoint(config, checkpoint_dir=tts_dir, eval=True)
|
| 43 |
+
|
| 44 |
+
# Move the model to GPU (if available)
|
| 45 |
+
if torch.cuda.is_available():
|
| 46 |
+
model.cuda()
|
| 47 |
+
|
| 48 |
+
return model
|
| 49 |
|
| 50 |
# Translation class
|
| 51 |
+
class Translation:
|
| 52 |
def __init__(self, video_path, original_language, target_language):
|
| 53 |
self.video_path = video_path
|
| 54 |
self.original_language = original_language
|
| 55 |
self.target_language = target_language
|
| 56 |
+
self.model = initialize_xtts_model() # Initialize TTS model
|
| 57 |
|
| 58 |
def org_language_parameters(self, original_language):
|
| 59 |
language_codes = {'English': 'en', 'German': 'de', 'Italian': 'it', 'Spanish': 'es'}
|
|
|
|
| 91 |
return translation
|
| 92 |
|
| 93 |
def generate_audio(self, translated_text):
|
| 94 |
+
# Generate audio using the xtts model
|
| 95 |
+
config = XttsConfig()
|
| 96 |
+
config.load_json(os.path.join(os.getcwd(), 'xtts_v2', 'config.json'))
|
| 97 |
+
|
| 98 |
+
# Generate audio
|
| 99 |
+
synthesized_audio_path = "output_synth.wav"
|
| 100 |
+
outputs = self.model.synthesize(
|
| 101 |
+
translated_text,
|
| 102 |
+
config,
|
| 103 |
+
speaker_wav='output_audio.wav',
|
| 104 |
+
gpt_cond_len=3,
|
| 105 |
+
language=self.tran_code,
|
| 106 |
+
)
|
| 107 |
+
|
| 108 |
+
# Save the output to file
|
| 109 |
+
with open(synthesized_audio_path, 'wb') as f:
|
| 110 |
+
f.write(outputs)
|
| 111 |
+
|
| 112 |
+
return synthesized_audio_path
|
| 113 |
|
| 114 |
def translate_video(self):
|
| 115 |
audio_path = self.extract_audio()
|
|
|
|
| 123 |
os.system(f"python inference.py --checkpoint_path 'checkpoints/wav2lip_gan.pth' --face {self.video_path} --audio {translated_audio_path} --outfile 'output_video.mp4'")
|
| 124 |
return 'output_video.mp4'
|
| 125 |
|
|
|
|
| 126 |
# Gradio Interface
|
| 127 |
def app(video_path, original_language, target_language):
|
| 128 |
+
translator = Translation(video_path, original_language, target_language)
|
| 129 |
video_file = translator.translate_video()
|
| 130 |
return video_file
|
| 131 |
|
|
|
|
| 139 |
outputs=gr.Video(label="Translated Video")
|
| 140 |
)
|
| 141 |
|
| 142 |
+
interface.launch()
|