AgentSkillsCN

voice-command-listener

捕捉麦克风输入并将其转录为文本指令，借助SoX进行录音，再利用OpenAI Whisper实现高精度转写，从而支持语音控制客服代理。

SKILL.md
--- frontmatter
name: voice-command-listener
description: Captures microphone input and transcribes it into text commands. Uses SoX for recording and OpenAI Whisper for high-accuracy transcription to control the agent via voice.
status: implemented

Voice Command Listener

This skill allows you to control the Gemini Ecosystem with your voice.

Capabilities

1. Voice Recording

  • •Uses sox (Sound eXchange) to capture audio from the default microphone.
  • •Records until silence is detected or a specific key is pressed (Ctrl+C).

2. Transcription (Whisper)

  • •Sends the recorded .wav or .mp3 file to the OpenAI Whisper API.
  • •Returns the transcribed text, ready to be piped into other skills.

Usage

  • •"Listen to my command." (Starts recording)
  • •"Transcribe this audio file and execute it as a prompt."

Prerequisites

  • •sox installed (brew install sox).
  • •OpenAI API Key configured in knowledge/personal/voice/config.json.

Knowledge Protocol

  • •Adheres to knowledge/tech-stack/macos/voice_control_guide.md.