Skip to content

Brief Mode

One-way explanation via TTS. The agent reads back a longer piece of content — decisions, summaries, findings — while the user listens. No response expected.

When to Use

  • Reading back analysis results: "Here's what I found in the codebase..."
  • Summarizing a plan: "The migration will happen in three phases..."
  • Explaining a decision: "I chose Redis over Memcached because..."

MCP Tool

Name: qa_voice_brief

Parameters

Parameter Type Required Default Description
message string Yes — The explanation or summary to speak (non-empty after trimming)
rate string No -10% Speech rate override (e.g., -20%, +0%)

Returns

{
  "content": [{ "type": "text", "text": "[brief] Explained: \"your message\"" }]
}

Errors

Same as Announce — edge-tts or audio player issues.

Behavior

Identical pipeline to announce (edge-tts -> audio player -> cleanup), but with a slower default rate optimized for longer content.

Speech Rate

Default: -10% (slower for comprehension of longer content).

Auto-adjusted for text length:

  • 300-599 chars: -15%
  • 600-999 chars: -20%
  • 1000+ chars: -25%

This makes brief mode noticeably slower than announce — intentional, since brief content is typically 2-5 sentences that the user needs to absorb.

Stop Signal

Same as all TTS modes:

touch /tmp/voicelayer-stop