Comprehensive MOSS Transcribe Diarize workflow for high-confidence multi-speaker ASR. Use when users need (1) timestamped transcription, (2) speaker-labeled...
Call this skill when users want:
scripts/transcribe.pysegments 生成:逐段文本、按说话人汇总、会后纪要moss-transcribe-diarizeaudio_data(URL 或 data URL)modelsampling_params(如 max_new_tokens, temperature)meta_info(可选)textmeta_infosegments(含时间戳、speaker、content)官方文档入口:
https://studio.mosi.cn/docs/moss-transcribe-diarize
# URL 音频
python scripts/transcribe.py \
--audio-url "https://example.com/audio.mp3" \
--api-key "$MOSS_API_KEY" \
--out result.json
# 本地文件(自动转 data URL)
python scripts/transcribe.py \
--file "/path/to/meeting.mp4" \
--api-key "$MOSS_API_KEY" \
--out result.json
默认 endpoint:https://studio.mosi.cn/v1/audio/transcriptions
如果你的环境 endpoint 不同,用参数覆盖:
--endpoint "https://your-endpoint"
*.segments.txt(逐段)*.by_speaker.txt(按说话人)ZIP package — ready to use