speakturbo-tts
by emzod
AI & LLMsBrief
Give your agent the ability to speak to you real-time.
Description
What This Skill Does
Adds real-time text-to-speech output to an agent, using a local daemon that streams audio with ~90ms latency after warmup. Ships with 8 built-in voices and outputs 24kHz mono audio. Not designed for voice cloning or emotion tags.
At ~90ms to first sound it's significantly faster than alternatives like the speak skill (~4-8s), making it practical for real-time conversational feedback.
When to Use It
- Hearing agent status updates hands-free while coding
- Getting spoken confirmations after long-running tasks finish
- Running voice-narrated demos without a cloud TTS service
- Saving agent responses as WAV files for later playback
- Quick audio feedback in terminal workflows
Example Workflow
User asks agent to summarize a pull request and read it aloud
- 1
Agent fetches and summarizes the pull request text
- 2
Agent calls speakturbo with the summary string
- 3
Daemon receives the text over HTTP on port 7125
- 4
Audio streams back and playback begins within ~90ms
Summary plays through speakers; agent reports playback complete
Installation
clawhub install emzod/speakturbo-ttsRun the CLI command, or paste the prompt into Claude Code, Cursor, Codex or whichever coding agent you use.