feat(xiaoyuzhou): Chinese punctuation prompt + optional --polish flag (#291)

Adds a punctuation-aware Whisper prompt and an optional --polish step (free Llama 3.3 70B on Groq) for Chinese podcast transcripts. Maintainer follow-up on the branch: replaced ASCII quotes that bash swallowed inside the prompt (hexdump-verified), added the same prompt to the 429-retry call, and documented --polish as optional. bash -n passes.
2026-06-10 14:56:38 +08:00
parent a827452881
commit 5b07d95e7e
2 changed files with 111 additions and 7 deletions
@@ -71,13 +71,15 @@ bili rank -n 10

 ## 小宇宙播客 / Xiaoyuzhou Podcast

-### 转录单集播客
+### 转录单集播客（可选 --polish 增强标点）

 ```bash
-# 输出 Markdown 文件到 /tmp/
-~/.agent-reach/tools/xiaoyuzhou/transcribe.sh "https://www.xiaoyuzhoufm.com/episode/EPISODE_ID"
+# 输出 Markdown 文件到 /tmp/。--polish 让 Llama 3.3 70B 给文稿补中文标点+合理分段
+~/.agent-reach/tools/xiaoyuzhou/transcribe.sh --polish "https://www.xiaoyuzhoufm.com/episode/EPISODE_ID"
 ```

+> 转写 prompt 已要求 Whisper 输出中文标点；若标点效果仍不理想，可加 `--polish` 用 Groq 上免费的 Llama 3.3 70B 补标点+合理分段（9 分钟播客约多 ~7 秒）。每次转写多一轮 LLM 调用，按需使用。
+
 ### 前置要求

 1. **ffmpeg**: `brew install ffmpeg`