TTS 前剥离 Markdown，避免语音读出格式符号噪音（PR #13237）

用于“语音播报把 `**粗体**`、链接括号、代码块标记都读出来”的场景：在 TTS 前统一清洗 markdown，提升可听性并减少误读。

GITHUBDiscovered 2026-02-13Author danielwanwx

Prerequisites

OpenClaw runtime includes PR #13237 (or equivalent strip-markdown implementation).
Your workflow uses TTS output for assistant replies or alerts.

Steps

Upgrade to a version including PR #13237 and restart gateway.
Prepare mixed-format test prompts (bold, lists, links, inline code, fenced code).
Trigger TTS generation before/after upgrade and compare spoken output quality.
If your environment has custom pre-processors, ensure markdown cleanup order does not remove critical semantics.

Commands

openclaw gateway status

openclaw gateway restart

openclaw status

Verify

TTS output no longer reads formatting artifacts while preserving sentence meaning and link context.

Caveats

Edge cases (math/markdown tables/custom tags) may still need additional handling（需验证）.
Over-aggressive stripping can drop useful structure; regression samples should be kept.

Source attribution

This tip is aggregated from community/public sources and preserved with attribution.