Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -33,33 +33,6 @@ languages = {
|
|
33 |
qa_pipeline = pipeline(task="question-answering", model="deepset/roberta-base-squad2")
|
34 |
|
35 |
|
36 |
-
# Function to download audio from YouTube
|
37 |
-
def download_audio_from_youtube(youtube_url, output_path="downloaded_audio.mp3"):
|
38 |
-
ydl_opts = {
|
39 |
-
'format': 'bestaudio/best',
|
40 |
-
'outtmpl': 'temp_audio.%(ext)s',
|
41 |
-
'postprocessors': [{
|
42 |
-
'key': 'FFmpegExtractAudio',
|
43 |
-
'preferredcodec': 'mp3',
|
44 |
-
'preferredquality': '192',
|
45 |
-
}],
|
46 |
-
'quiet': True,
|
47 |
-
'no_warnings': True,
|
48 |
-
}
|
49 |
-
|
50 |
-
try:
|
51 |
-
command = [
|
52 |
-
"yt-dlp",
|
53 |
-
"-x", # extract audio only
|
54 |
-
"--audio-format", "mp3", # specify mp3 format
|
55 |
-
"-o", output_path, # specify output path
|
56 |
-
youtube_url # YouTube URL
|
57 |
-
]
|
58 |
-
subprocess.run(command, check=True, capture_output=True)
|
59 |
-
return output_path
|
60 |
-
except Exception as e:
|
61 |
-
return f"Error downloading audio: {e}"
|
62 |
-
|
63 |
# Function to extract audio from video
|
64 |
def extract_audio_from_video(video_file, output_audio="extracted_audio.mp3"):
|
65 |
try:
|
@@ -75,23 +48,18 @@ languageG = None
|
|
75 |
|
76 |
def content_input_update(content_type):
|
77 |
visibility_map = {
|
78 |
-
"Audio Upload": (True, False
|
79 |
-
"Video Upload": (False,
|
80 |
-
"YouTube Link": (False, True, False),
|
81 |
}
|
82 |
-
visible_audio,
|
83 |
return (
|
84 |
gr.update(visible=visible_audio),
|
85 |
-
gr.update(visible=visible_youtube),
|
86 |
gr.update(visible=visible_video)
|
87 |
)
|
88 |
|
89 |
def transcribe_content(content_type, audio_path, youtube_link, video):
|
90 |
if content_type == "Audio Upload" and audio_path:
|
91 |
return whispermodel.transcribe(audio_path)["text"]
|
92 |
-
elif content_type == "YouTube Link" and youtube_link:
|
93 |
-
audio_file = download_audio_from_youtube(youtube_link)
|
94 |
-
return whispermodel.transcribe(audio_file)["text"]
|
95 |
elif content_type == "Video Upload" and video:
|
96 |
audio_file = extract_audio_from_video(video.name)
|
97 |
return whispermodel.transcribe(audio_file)["text"]
|
@@ -128,7 +96,7 @@ def create_audio_summary(summary, language):
|
|
128 |
return audio_path
|
129 |
return None
|
130 |
|
131 |
-
def main(content_type, audio_path,
|
132 |
global transcription, languageG
|
133 |
languageG = language
|
134 |
|
@@ -159,26 +127,24 @@ with gr.Blocks() as demo:
|
|
159 |
)
|
160 |
|
161 |
content_type = gr.Radio(
|
162 |
-
choices=["Audio Upload", "Video Upload"
|
163 |
label="Select Content Type",
|
164 |
value="Audio Upload"
|
165 |
)
|
166 |
|
167 |
file_input = gr.Audio(label="Upload an Audio File", visible=True, type="filepath")
|
168 |
-
youtube_input = gr.Textbox(label="Enter YouTube Link", visible=False, placeholder="https://www.youtube.com/watch?v=example")
|
169 |
video_input = gr.File(label="Upload a Video", visible=False, type="filepath")
|
170 |
|
171 |
language = gr.Radio(choices=["Arabic", "English"], label="Preferred Language", value="English")
|
172 |
summarize = gr.Checkbox(label="Summarize the content?")
|
173 |
|
174 |
examples = [
|
175 |
-
["Audio Upload", "audio-example.mp3",
|
176 |
-
["Video Upload", None,
|
177 |
-
["YouTube Link", None, "https://www.youtube.com/watch?v=J4RqCSD--Dg&ab_channel=LearnFree", None, "English", False, True, 2]
|
178 |
]
|
179 |
gr.Examples(
|
180 |
examples=examples,
|
181 |
-
inputs=[content_type, file_input,
|
182 |
label="Try These Examples"
|
183 |
)
|
184 |
|
|
|
33 |
qa_pipeline = pipeline(task="question-answering", model="deepset/roberta-base-squad2")
|
34 |
|
35 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
36 |
# Function to extract audio from video
|
37 |
def extract_audio_from_video(video_file, output_audio="extracted_audio.mp3"):
|
38 |
try:
|
|
|
48 |
|
49 |
def content_input_update(content_type):
|
50 |
visibility_map = {
|
51 |
+
"Audio Upload": (True, False),
|
52 |
+
"Video Upload": (False, True),
|
|
|
53 |
}
|
54 |
+
visible_audio, visible_video = visibility_map.get(content_type, (False, False))
|
55 |
return (
|
56 |
gr.update(visible=visible_audio),
|
|
|
57 |
gr.update(visible=visible_video)
|
58 |
)
|
59 |
|
60 |
def transcribe_content(content_type, audio_path, youtube_link, video):
|
61 |
if content_type == "Audio Upload" and audio_path:
|
62 |
return whispermodel.transcribe(audio_path)["text"]
|
|
|
|
|
|
|
63 |
elif content_type == "Video Upload" and video:
|
64 |
audio_file = extract_audio_from_video(video.name)
|
65 |
return whispermodel.transcribe(audio_file)["text"]
|
|
|
96 |
return audio_path
|
97 |
return None
|
98 |
|
99 |
+
def main(content_type, audio_path, video, language, summarize):
|
100 |
global transcription, languageG
|
101 |
languageG = language
|
102 |
|
|
|
127 |
)
|
128 |
|
129 |
content_type = gr.Radio(
|
130 |
+
choices=["Audio Upload", "Video Upload"],
|
131 |
label="Select Content Type",
|
132 |
value="Audio Upload"
|
133 |
)
|
134 |
|
135 |
file_input = gr.Audio(label="Upload an Audio File", visible=True, type="filepath")
|
|
|
136 |
video_input = gr.File(label="Upload a Video", visible=False, type="filepath")
|
137 |
|
138 |
language = gr.Radio(choices=["Arabic", "English"], label="Preferred Language", value="English")
|
139 |
summarize = gr.Checkbox(label="Summarize the content?")
|
140 |
|
141 |
examples = [
|
142 |
+
["Audio Upload", "audio-example.mp3", None, "English", True, True, 5],
|
143 |
+
["Video Upload", None, "video-example.mp4", "Arabic", True, False, 3],
|
|
|
144 |
]
|
145 |
gr.Examples(
|
146 |
examples=examples,
|
147 |
+
inputs=[content_type, file_input, video_input, language, summarize],
|
148 |
label="Try These Examples"
|
149 |
)
|
150 |
|