sharepoint-transcribe/sp-transcribe-service/lambdas/transcribe_start.py

"""Start Amazon Transcribe job with language selection."""
import os
import boto3
from shared import S3_BUCKET, update_job

transcribe = boto3.client('transcribe')

AUTO_DETECT_LANGUAGES = ['zh-CN', 'zh-TW', 'yue-CN', 'en-US']

# Amazon Transcribe 语言代码映射
LANG_MAP = {
    'zh-CN': 'zh-CN',
    'zh-TW': 'zh-TW',
    'zh-HK': 'zh-HK',
    'en-US': 'en-US',
}


def handler(event, context):
    job_id = event['job_id']
    audio_s3_key = event['audio_s3_key']
    language = event.get('language', 'auto')
    transcribe_job = f"sp-{job_id}"

    update_job(job_id, status='TRANSCRIBING')

    params = {
        'TranscriptionJobName': transcribe_job,
        'Media': {'MediaFileUri': f"s3://{S3_BUCKET}/{audio_s3_key}"},
        'OutputBucketName': S3_BUCKET,
        'OutputKey': f"jobs/{job_id}/transcribe-output.json",
        'Settings': {
            'ShowSpeakerLabels': True,
            'MaxSpeakerLabels': 10,
        },
    }

    if language == 'auto':
        # 自动语言识别，提供候选语言列表
        params['IdentifyLanguage'] = True
        params['LanguageOptions'] = AUTO_DETECT_LANGUAGES
    else:
        # 指定语言
        params['LanguageCode'] = LANG_MAP.get(language, language)

    transcribe.start_transcription_job(**params)

    return {**event, 'transcribe_job': transcribe_job}