🧪 Skills
Moss Transcribe Diarize
Comprehensive MOSS Transcribe Diarize workflow for high-confidence multi-speaker ASR. Use when users need (1) timestamped transcription, (2) speaker-labeled...
v0.1.1
Description
name: moss-transcribe-diarize description: Comprehensive MOSS Transcribe Diarize workflow for high-confidence multi-speaker ASR. Use when users need (1) timestamped transcription, (2) speaker-labeled segments/diarization, (3) meeting or interview transcript extraction, or (4) ASR from URL/Base64/local audio-video files with structured post-processing outputs (raw JSON, segment timeline, and by-speaker summary).
MOSS Transcribe Diarize Skill
Call this skill when users want:
- 多人语音转写(带说话人)
- 带时间戳的会议纪要原文
- 从音视频 URL / 本地文件做 ASR + diarization
Quick workflow
- 准备音频来源(URL / 本地文件 / Base64)
- 调用
scripts/transcribe.py - 用
segments生成:逐段文本、按说话人汇总、会后纪要
API assumptions (from docs page)
- 模型名固定:
moss-transcribe-diarize - 请求体核心字段:
audio_data(URL 或 data URL)modelsampling_params(如max_new_tokens,temperature)meta_info(可选)
- 返回中重点看:
textmeta_infosegments(含时间戳、speaker、content)
官方文档入口:
https://studio.mosi.cn/docs/moss-transcribe-diarize
Run
# 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
Endpoint
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"
Output handling
- 原始结果保存为 JSON
- 脚本会额外导出:
*.segments.txt(逐段)*.by_speaker.txt(按说话人)
Reviews (0)
Sign in to write a review.
No reviews yet. Be the first to review!
Comments (0)
No comments yet. Be the first to share your thoughts!