app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,60 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
import whisper
|
| 3 |
+
from whisper.utils import write_vtt
|
| 4 |
+
import requests
|
| 5 |
+
import os
|
| 6 |
+
|
| 7 |
+
model = whisper.load_model("medium")
|
| 8 |
+
|
| 9 |
+
def inference(link):
|
| 10 |
+
content = requests.get(link)
|
| 11 |
+
podcast_url = re.findall("(?P<url>\;https?://[^\s]+)", content.text)[0].split(';')[1]
|
| 12 |
+
print(podcast_url)
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
download = requests.get(podcast_url)
|
| 16 |
+
|
| 17 |
+
with open('podcast.mp3', 'wb') as f:
|
| 18 |
+
f.write(download.content)
|
| 19 |
+
|
| 20 |
+
result = model.transcribe('podcast.mp3')
|
| 21 |
+
|
| 22 |
+
with open('sub.vtt', "w") as txt:
|
| 23 |
+
write_vtt(result["segments"], file=txt)
|
| 24 |
+
|
| 25 |
+
return (result['text'], 'sub.vtt')
|
| 26 |
+
|
| 27 |
+
|
| 28 |
+
|
| 29 |
+
|
| 30 |
+
title="PodScript"
|
| 31 |
+
description="Get Podcast Transcript"
|
| 32 |
+
block = gr.Blocks()
|
| 33 |
+
|
| 34 |
+
with block:
|
| 35 |
+
gr.HTML(
|
| 36 |
+
""" <center>
|
| 37 |
+
<h1>PodScript</h1>
|
| 38 |
+
<img src = 'https://external-content.duckduckgo.com/iu/?u=https%3A%2F%2Fwww.newbreedmarketing.com%2Fhs-fs%2Fhubfs%2Fshutterstock_1125707303.jpg%3Fwidth%3D5000%26name%3Dshutterstock_1125707303.jpg&f=1&nofb=1&ipt=0ba3d9b639d63b0b737ea63cd81b241cc47f46cb519ac4ca18fd3ce6fc1376ad&ipo=images' width = '50%'></img>
|
| 39 |
+
</center>
|
| 40 |
+
"""
|
| 41 |
+
)
|
| 42 |
+
with gr.Group():
|
| 43 |
+
with gr.Box():
|
| 44 |
+
|
| 45 |
+
link = gr.Textbox(label="Google Podcasts Link")
|
| 46 |
+
|
| 47 |
+
with gr.Row().style(mobile_collapse=False, equal_height=True):
|
| 48 |
+
btn = gr.Button("Get PodScript 🪄")
|
| 49 |
+
|
| 50 |
+
text = gr.Textbox(
|
| 51 |
+
label="PodScript",
|
| 52 |
+
placeholder="PodScript Output",
|
| 53 |
+
lines=5)
|
| 54 |
+
|
| 55 |
+
file = gr.File()
|
| 56 |
+
|
| 57 |
+
|
| 58 |
+
btn.click(inference, inputs=[link], outputs=[text,file])
|
| 59 |
+
|
| 60 |
+
block.launch(debug=True)
|