Create app.py

app.py

@@ -0,0 +1,80 @@
+import gradio as gr
+import json
+import spacy
+import re
+import string
+import pandas as pd
+import os
+import requests
+
+nlp = spacy.load("en_core_web_sm")
+nlp.add_pipe('sentencizer')
+
+
+
+def read_gs(sheet_url):
+    s_url = sheet_url.replace('/edit#gid=', '/export?format=csv&gid=')
+    df = pd.read_csv(s_url)
+    return df
+
+
+def download_and_save_file(URL, audio_dir):
+    headers = {
+        'user-agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/72.0.3626.121 Safari/537.36',
+        'accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8',
+        'referer': 'https://www.google.com/',
+        'accept-encoding': 'gzip, deflate, br',
+        'accept-language': 'en-US,en;q=0.9,',
+        'cookie': 'prov=6bb44cc9-dfe4-1b95-a65d-5250b3b4c9fb; _ga=GA1.2.1363624981.1550767314; __qca=P0-1074700243-1550767314392; notice-ctt=4%3B1550784035760; _gid=GA1.2.1415061800.1552935051; acct=t=4CnQ70qSwPMzOe6jigQlAR28TSW%2fMxzx&s=32zlYt1%2b3TBwWVaCHxH%2bl5aDhLjmq4Xr',
+    }
+    doc = requests.get(URL, headers=headers)
+    file_name = URL.split('/')[-1].split('?')[0]
+    audio_path = f'{audio_dir}/{file_name}'
+    with open(audio_path, 'wb') as f:
+        f.write(doc.content)
+    return audio_path
+
+def select_samples():
+
+    df = read_gs('https://docs.google.com/spreadsheets/d/17QG4puJRXN8V5froIv8YrJIMsns0GTt4/edit#gid=1020901598')
+    audio_dir = 'AUDIO'
+    os.makedirs(audio_dir, exist_ok=True)
+    df = df.sample(1)
+
+
+    audio_url = df.url.values[0]
+
+    audio_path = download_and_save_file(audio_url, audio_dir)
+    return audio_path, df['text'].values[0]
+
+
+title = '🎵 Annotate audio'
+description = '''Choose a sentence that describes audio the best if there's no such sentence please choose `No audio description`'''
+
+audio_path, full_text = select_samples()
+full_text = full_text.translate(str.maketrans('', '', string.punctuation))
+sents = [re.sub(r'###audio###\d###', '', s.text) for s in nlp(full_text).sents]
+sents.append('No audio description')
+
+def audio_demo(text, audio, audio_id):
+
+  with open('data.json', 'w') as f:
+    data = {
+        'audio':audio_id,
+        'text':text
+    }
+    json.dump(data, f)
+    return 'success!'
+
+
+iface = gr.Interface(
+    audio_demo,  
+    inputs=[gr.Dropdown(sents, label='audio description'), gr.Audio(audio_path, type="filepath"), gr.Textbox(value=audio_path, visible=False)],
+    outputs=[gr.Textbox(label="output")],
+    allow_flagging="never",
+    title=title,
+    description=description,
+    )
+
+if __name__ == "__main__":
+  iface.launch(show_error=True, debug=True)