Spaces:
Runtime error
Runtime error
PhenixNova
commited on
Commit
•
cec1f0b
1
Parent(s):
f881b76
Upload app.py
Browse files
app.py
ADDED
@@ -0,0 +1,75 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import os
|
3 |
+
import subprocess
|
4 |
+
import whisper
|
5 |
+
from whisper.utils import write_vtt
|
6 |
+
|
7 |
+
model = whisper.load_model("medium")
|
8 |
+
|
9 |
+
def video2mp3(video_file, output_ext="mp3"):
|
10 |
+
filename, ext = os.path.splitext(video_file)
|
11 |
+
subprocess.call(["ffmpeg", "-y", "-i", video_file, f"{filename}.{output_ext}"],
|
12 |
+
stdout=subprocess.DEVNULL,
|
13 |
+
stderr=subprocess.STDOUT)
|
14 |
+
return f"{filename}.{output_ext}"
|
15 |
+
|
16 |
+
def transcribe_audio(file):
|
17 |
+
options = dict(task="transcribe", best_of=5, fp16 = False)
|
18 |
+
text = model.transcribe(file, **options)["text"]
|
19 |
+
return text.strip()
|
20 |
+
|
21 |
+
def translate_audio(file):
|
22 |
+
options = dict(task="translate", best_of=5, fp16 = False)
|
23 |
+
text = model.transcribe(file, **options)["text"]
|
24 |
+
return text.strip()
|
25 |
+
|
26 |
+
def translate_video(input_video):
|
27 |
+
audio_file = video2mp3(input_video)
|
28 |
+
|
29 |
+
options = dict(beam_size=5, best_of=5, fp16 = False)
|
30 |
+
translate_options = dict(task="translate", **options)
|
31 |
+
result = model.transcribe(audio_file,**translate_options)
|
32 |
+
|
33 |
+
output_dir = '/content/'
|
34 |
+
audio_path = audio_file.split(".")[0]
|
35 |
+
|
36 |
+
with open(os.path.join(output_dir, audio_path + ".vtt"), "w") as vtt:
|
37 |
+
write_vtt(result["segments"], file=vtt)
|
38 |
+
|
39 |
+
subtitle = audio_path + ".vtt"
|
40 |
+
output_video = audio_path + "_subtitled.mp4"
|
41 |
+
|
42 |
+
os.system(f"ffmpeg -i {input_video} -vf subtitles={subtitle} {output_video}")
|
43 |
+
|
44 |
+
return output_video
|
45 |
+
|
46 |
+
block = gr.Blocks()
|
47 |
+
|
48 |
+
with block:
|
49 |
+
|
50 |
+
with gr.Group():
|
51 |
+
with gr.Box():
|
52 |
+
with gr.Row().style():
|
53 |
+
inp_video = gr.Video(
|
54 |
+
label="Input Video",
|
55 |
+
type="filepath",
|
56 |
+
mirror_webcam = False
|
57 |
+
)
|
58 |
+
op_video = gr.Video()
|
59 |
+
btn = gr.Button("Generate Subtitle Video")
|
60 |
+
btn.click(translate_video, inputs=[inp_video], outputs=[op_video])
|
61 |
+
with gr.Group():
|
62 |
+
audio = gr.Audio(
|
63 |
+
show_label=False,
|
64 |
+
source="microphone",
|
65 |
+
type="filepath"
|
66 |
+
)
|
67 |
+
with gr.Box():
|
68 |
+
with gr.Row().style(equal_height=True):
|
69 |
+
transcribe_button = gr.Button("Transcribe")
|
70 |
+
translate_button = gr.Button("Translate")
|
71 |
+
textbox = gr.Textbox(show_label=False)
|
72 |
+
transcribe_button.click(transcribe_audio, inputs=[audio], outputs=[textbox])
|
73 |
+
translate_button.click(translate_audio, inputs=[audio], outputs=[textbox])
|
74 |
+
|
75 |
+
block.launch(debug = True)
|