Edge Tts Global

Use the globally installed edge-tts command to generate Chinese or multilingual text-to-speech audio on this machine. Use when the user wants text converted to speech, wants a natural online voice quickly, wants Telegram voice-bubble style audio, or wants help choosing Edge TTS voices and commands.

Use this skill when the user wants fast text-to-speech with the globally installed edge-tts command.

What is available

Global commands:

• edge-tts
• edge-playback

Bundled helper script:

• scripts/tts_edge.py

Preferred Chinese test voice:

• zh-CN-XiaoxiaoNeural

Fast workflow

1. Put output files in the current caller workspace temp/ directory.
2. Prefer the bundled Python script scripts/tts_edge.py for standard generation.
3. If needed, call edge-tts directly.
4. If sending back to Telegram and the user wants a voice bubble, send the audio with message and set asVoice=true.
5. If the user wants a normal file/audio attachment, send it normally.
6. After successful sending, immediately delete the generated temporary file.

Preferred script usage

Generate audio to the current caller workspace temp/ directory automatically:

bash

python3 /data/claw/shared/skills/edge-tts-global/scripts/tts_edge.py -t "老板，你好，我是你的小助理。"

Generate audio with explicit relative path under the current caller workspace:

bash

python3 /data/claw/shared/skills/edge-tts-global/scripts/tts_edge.py -t "老板，你好，我是你的小助理。" -o ./temp/out.mp3

Delete temporary file after successful sending:

bash

python3 /data/claw/shared/skills/edge-tts-global/scripts/tts_edge.py cleanup ./temp/out.mp3

Delete both audio and subtitles:

bash

python3 /data/claw/shared/skills/edge-tts-global/scripts/tts_edge.py cleanup ./temp/out.mp3 ./temp/out.vtt

Change voice explicitly:

bash

python3 /data/claw/shared/skills/edge-tts-global/scripts/tts_edge.py -t "你好" -v zh-CN-XiaoxiaoNeural -o ./temp/out.mp3

Generate subtitles too:

bash

python3 /data/claw/shared/skills/edge-tts-global/scripts/tts_edge.py -t "你好" -o ./temp/out.mp3 --subs ./temp/out.vtt

Direct command patterns

Generate MP3

bash

mkdir -p ./temp
edge-tts --voice zh-CN-XiaoxiaoNeural --text "老板，你好，我是你的小助理。" --write-media ./temp/out.mp3

List voices

bash

edge-tts --list-voices

Telegram sending rule

If the user asks for “气泡语音”, “语音条”, or a voice-message style result:

• generate the file first
• send with message
• set asVoice=true
• after successful sending, delete the temporary file immediately

If the user asks for a normal audio file:

• send as a regular attachment
• do not set asVoice=true
• after successful sending, delete the temporary file immediately

Cleanup guidance

• Temporary outputs belong in the current caller workspace temp/ directory
• Sending succeeds → delete the generated temporary files immediately
• Prefer python3 scripts/tts_edge.py cleanup <file> [subtitle] for deterministic cleanup
• Do not remove the global edge-tts install unless the user explicitly asks