File size: 3,103 Bytes
5f4a5e8 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 | #!/usr/bin/env python3
"""Preenche as transcricoes vazias do metadata.csv usando Whisper.
Uso (na pasta fala_pb, na sua maquina com whisper instalado):
python transcribe_missing.py # modelo small
python transcribe_missing.py --model medium
Requer: pip install openai-whisper (voce ja usou no process_dataset.py)
Retomavel: salva progresso em transcricoes_parciais.csv a cada arquivo;
se interromper, rode de novo que continua de onde parou.
"""
import argparse
import csv
import os
import re
import sys
import unicodedata
csv.field_size_limit(10_000_000)
BASE = os.path.dirname(os.path.abspath(__file__))
META = os.path.join(BASE, 'metadata.csv')
PARTIAL = os.path.join(BASE, 'transcricoes_parciais.csv')
def norm_text(t):
if not t:
return ''
t = t.replace('...', ' ').replace('…', ' ')
t = re.sub(r'[^\w\sÀ-ÿ]', ' ', t, flags=re.UNICODE)
return re.sub(r'\s+', ' ', t).strip().lower()
def main():
ap = argparse.ArgumentParser()
ap.add_argument('--model', default='small')
args = ap.parse_args()
rows = list(csv.DictReader(open(META, encoding='utf-8')))
pending = [r for r in rows if not r['text'].strip()]
print(f'{len(pending)} audios sem transcricao')
if not pending:
return
done = {}
if os.path.exists(PARTIAL):
for r in csv.DictReader(open(PARTIAL, encoding='utf-8')):
done[r['id']] = r['text']
print(f'{len(done)} ja transcritos anteriormente')
todo = [r for r in pending if r['id'] not in done]
if todo:
import whisper
model = whisper.load_model(args.model)
new_file = not os.path.exists(PARTIAL)
with open(PARTIAL, 'a', newline='', encoding='utf-8') as pf:
w = csv.writer(pf)
if new_file:
w.writerow(['id', 'text'])
for i, r in enumerate(todo, 1):
path = os.path.join(BASE, r['file_name'].replace('/', os.sep))
try:
res = model.transcribe(path, language='pt')
text = res['text'].strip()
except Exception as e:
print(f'ERRO {r["id"]}: {e}')
text = ''
done[r['id']] = text
w.writerow([r['id'], text])
pf.flush()
print(f'[{i}/{len(todo)}] {r["id"]}: {text[:70]}')
# aplica no metadata.csv
for r in rows:
if r['id'] in done and done[r['id']]:
r['text'] = done[r['id']]
r['normalized_text'] = norm_text(done[r['id']])
r['notes'] = (r['notes'].replace(
'transcricao pendente (transcribe_missing.py)',
'transcricao automatica (Whisper, revisao manual recomendada)'))
tmp = META + '.tmp'
with open(tmp, 'w', newline='', encoding='utf-8') as f:
w = csv.DictWriter(f, fieldnames=rows[0].keys())
w.writeheader()
w.writerows(rows)
os.replace(tmp, META)
print(f'metadata.csv atualizado. Pode apagar {os.path.basename(PARTIAL)}.')
if __name__ == '__main__':
main()
|