marianna13/annotate-audio

1 2import gradio as gr3import json4import re5import string6import pandas as pd7import os8import requests9from textwrap import wrap10import uuid11import gspread12import ast13 14 15 16def download_and_save_file(URL, audio_dir):17    headers = {18        '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',19        'accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8',20        'referer': 'https://www.google.com/',21        'accept-encoding': 'gzip, deflate, br',22        'accept-language': 'en-US,en;q=0.9,',23        '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',24    }25    doc = requests.get(URL, headers=headers)26    file_name = URL.split('/')[-1].split('?')[0]27    audio_path = f'{audio_dir}/{file_name}'28    with open(audio_path, 'wb') as f:29        f.write(doc.content)  30    return audio_path31 32 33 34credentials = os.environ['CREDENTIALS']35data = json.loads(credentials, strict=False)36with open('credentials.json', 'w') as f:37    json.dump(data, f)38 39 40 41 42title = '🎵 Annotate audio'43description = '''Choose a sentence (or sentences) that describes audio the best.'''44 45audio_dir = 'AUDIO'46os.makedirs(audio_dir, exist_ok=True)47 48def sample_df():49 50    gc = gspread.service_account(filename='credentials.json')51    sh = gc.open('Annotated CC Audio')52    worksheet = sh.sheet153    df = pd.DataFrame(worksheet.get_all_records())54    sample_df = df[df['caption']==''].sample(1)55 56    audio_url, audio_meta, page_title, img_metadata, sibling_elems = sample_df[['audio_url', 'audio_meta', 'page_title', 'imgs_metadata', 'sibling_elems']].values[0]57    audio_path = download_and_save_file(audio_url, audio_dir)58    sibling_elems = ast.literal_eval(sibling_elems)59    sibling_elems = [s.replace('\n', '') for s in sibling_elems]60    sibling_elems = ["\n".join(wrap(s)) for s in sibling_elems if len(s) > 0]61    sibling_elems = list(set(sibling_elems))62    img_metadata = ast.literal_eval(img_metadata)63    if len(img_metadata) > 0:64      img_metadata = [[f'{k}: {meta[k]}' for k in meta] for meta in img_metadata]65    audio_meta = ast.literal_eval(audio_meta).get('tags', None)66    if audio_meta:67      audio_meta = [f'{k}: {audio_meta[k]}' for k in audio_meta.keys() if k.lower() in ['title', 'album', 'artist', 'genre', 'date', 'language']]68      audio_meta = '; '.join(audio_meta)69    return audio_path, audio_url, sibling_elems, audio_meta, page_title, df, worksheet70 71def audio_demo(siblings, page_title, audio_meta, audio, annotator, audio_url):72    annotator = annotator if annotator else str(uuid.uuid4())73    siblings.extend(page_title)74    siblings.extend(audio_meta)75    siblings = [s for s in siblings if s!=[]]76    cap = '\n'.join(siblings)77    df['caption'].loc[df['audio_url'] == audio_url] = cap78    df['annotator'].loc[df['audio_url'] == audio_url] = annotator79    worksheet.update([df.columns.values.tolist()] + df.values.tolist())80    return 'success!'81 82 83if __name__ == "__main__":84    audio_path, audio_url, sibling_elems, audio_meta, page_title, df, worksheet = sample_df()85 86    iface = gr.Interface(87        audio_demo,  88        inputs=[89            gr.CheckboxGroup(sibling_elems, label='sibling elements text'), 90            gr.CheckboxGroup(label='page title', choices=[page_title]), 91            gr.CheckboxGroup([audio_meta], label='audio metadata'), 92            gr.Audio(audio_path, type="filepath", interactive=False), 93            gr.Textbox(label='please enter your name'), 94            gr.Textbox(value=audio_url, visible=False)95            ],96        outputs=[gr.Textbox(label="output")],97        allow_flagging="never",98        title=title,99        description=description,100        )101 102    iface.launch(show_error=True, debug=True)