def transcribe_document(
document: Document,
aws_bucket_name: str,
lang: str,
):
filekey = document.filename
job_raw = transcribe(
filekey=filekey,
aws_bucket_name=aws_bucket_name,
lang=lang,
)
full_text = job_raw["results"]["transcripts"][0]["transcript"]
transcription = AWSTranscription(raw=job_raw, full_text=full_text)
document.words = []
for i, item in enumerate(job_raw["results"]["items"]):
if item["type"] == "punctuation":
continue
alternative = item["alternatives"][0]
assert "start_time" in item, item
word = Word(
word=alternative["content"],
order=i,
start_time=timedelta(seconds=float(item["start_time"])),
end_time=timedelta(seconds=float(item["end_time"])),
confidence=float(alternative["confidence"]),
)
document.words.append(word)
document.transcription = transcription
