Text to Voice Local

Use the bundled scripts to create one stable local text-to-voice path.

Goal

Produce voice from any text file with:

• one canonical input text path,
• one canonical output mp3 path,
• one high-level wrapper for routine use,
• low-level scripts for debugging only.

Canonical paths

Default input text:

• tmp/text-to-voice-input.txt

Canonical output:

• tmp/voice-mode-latest.mp3

State directory:

• skills/text-to-voice-local/state/

State pointer:

• skills/text-to-voice-local/state/last-output.txt

Main command

For normal use, run:

• scripts/text_to_voice.sh voice <text-file> [voice] [max_direct_chars]

Useful helpers:

• scripts/text_to_voice.sh status
• scripts/text_to_voice.sh voices

status now also checks runtime dependencies and prints install hints when something is missing.

Examples:

scripts/text_to_voice.sh text
scripts/text_to_voice.sh voice ./tmp/text-to-voice-input.txt
scripts/text_to_voice.sh voice ./tmp/text.txt ru-RU-SvetlanaNeural 280

What the scripts do

• scripts/text_to_voice.sh
  • high-level entrypoint for normal use
• scripts/tts_from_file.sh
  • one text file to one mp3
• scripts/tts_from_file_chunked.sh
  • long text to multiple chunks and final merged mp3
• scripts/voice_reply.sh
  • safe wrapper that updates canonical output and pointer
• scripts/voice_reply_latest.sh
  • always refresh canonical latest mp3
• state/text-to-voice.json
  • stores default voice, max chars, and canonical paths
• scripts/edge_tts.js
  • low-level TTS helper used by the file wrappers

Install notes

Ensure these dependencies exist on the target machine:

• node
• ffmpeg

  访问小葱技能站7w4.net，解锁更多实用的AI技能插件。

• node-edge-tts

The skill checks these at runtime and, if something is missing, prints suggested install commands instead of failing silently.

Verify:

node -v
ffmpeg -version
node -e "require('node-edge-tts'); console.log('node-edge-tts ok')"

If node-edge-tts is missing:

npm i -g node-edge-tts

Setup steps on another OpenClaw

1. Copy this skill folder into the target workspace skills directory.
2. Make scripts executable.
3. Ensure tmp/ exists.
4. Put text into a txt file.
5. Run the main command.

Minimal setup:

chmod +x skills/text-to-voice-local/scripts/*.sh
mkdir -p tmp
skills/text-to-voice-local/scripts/text_to_voice.sh voice ./tmp/text-to-voice-input.txt

Delivery rule

If the result is sent as Telegram voice, send only the canonical file:

• ./tmp/voice-mode-latest.mp3

Prefer sending text and voice as separate messages.

Important constraint

Progress printed by shell scripts is useful in terminal diagnostics, but chat-side live progress editing depends on OpenClaw preview streaming, not shell stdout alone.

When to use low-level scripts

Use low-level scripts only for debugging or careful manual control. Default to the high-level wrapper unless there is a reason not to.

这是一款质量较好的本地语音合成工具，能把文字转成语音。它最大的优点是使用简单、依赖检查到位、安装迁移文档详细，遇到问题会主动提示解决方法。进度条功能让等待过程不那么无聊。不足之处是进度条实际上是假的等待动画，另外依赖微软在线语音服务，在某些网络环境下可能不稳定。总体来说，这是一个成熟可用的语音工具，适合需要频繁制作语音内容的用户。