Commands

Soundboard Plugin exposes 5 slash commands for quick access and a full MCP tool surface for programmatic control.

Slash commands

Command	What it does
/soundboard:speak	General TTS with emotion detection and SSML support. Accepts plain text, SSML-lite tags, or dialogue markup.
/soundboard:narrate	Code walkthrough narration with adaptive pacing. Pass a file path and the plugin reads, chunks, and speaks the code with context-aware speed.
/soundboard:notify	Spoken workflow notifications. Designed for build, test, and deploy events. Uses the announcer voice preset by default.
/soundboard:voices	List all available voices and presets. Shows voice ID, gender, style, and default emotion routing.
/soundboard:voice-status	Engine health check. Reports backend info, active voice, queue depth, enforced limits, and any errors.

MCP tool surface

These tools are available when the voice engine is running. Claude Code calls them automatically based on context, or you can invoke them through the plugin.

voice.speak

Synthesise speech from text. Supports plain text, SSML-lite tags, and emotion hints.

Parameters:

• text (string, required) — the text to speak.
• voice (string, optional) — voice ID or preset name (e.g., am_fenrir, narrator).
• emotion (string, optional) — emotion hint (neutral, serious, friendly, professional, calm, joy, urgent, whisper).
• speed (number, optional) — playback speed multiplier (0.5 to 2.0).
• format (string, optional) — input format: plain (default) or ssml.
• chunking (string, optional) — chunking strategy: auto (default) or off.
• concat (boolean, optional) — if true, concatenate multi-chunk audio into a single WAV file.

voice.stream

Stream speech for low-latency playback. Chunks are synthesised and played as they arrive rather than waiting for the full text.

Parameters:

• text (string, required) — the text to stream.
• voice (string, optional) — voice ID override.

voice.dialogue

Multi-speaker dialogue synthesis. Parses dialogue markup with character names, stage directions, and speed modifiers. Auto-casts characters to voices.

Parameters:

• script (string, required) — dialogue script with Speaker: text lines. Supports [pause Xms] directives, # comments, and blank lines.
• cast (object, optional) — explicit speaker-to-voice mapping (keys are speaker names, values are voice IDs or preset names).
• speed (number, optional) — base speed multiplier for all lines (0.5 to 2.0).
• concat (boolean, optional) — if true, concatenate all line audio into a single WAV file (default: true).
• debug (boolean, optional) — if true, include cue sheet and parse warnings in response.

voice.inner_monologue

Ambient inner monologue. Rate-limited, automatically redacted, volatile. Requires VOICE_SOUNDBOARD_AMBIENT_ENABLED=1.

Parameters:

• text (string, required) — monologue text (max 500 chars, sensitive content auto-redacted).
• category (string, optional) — category tag: general (default), thinking, observation, or debug.

voice.interrupt

Stop all active audio immediately. Clears the playback queue.

Parameters: None.

voice.list_voices

List available voices and presets. Returns voice ID, gender, style, and emotion routing for each voice.

Parameters: None.

voice.status

Check engine health and capabilities. Returns backend info, active voice, queue state, and enforced limits.

Parameters: None.

voice.narrate

Code-aware narration with adaptive pacing. Designed for spoken code reviews and explanations.

Parameters:

• text (string, required) — the narration script to speak (plain-language walkthrough, not raw code).
• voice (string, optional) — voice ID or preset name (default: am_fenrir).
• speed (number, optional) — speed multiplier (default: 0.95 for narrator pace, range 0.5 to 2.0).

voice.workflow_notify

Speak a short workflow notification with context-aware emotion.

Parameters:

• event_type (string, required) — type of workflow event: build_success, build_fail, tests_pass, tests_fail, commit, pr_merged, or deploy.
• message (string, required) — short message to speak (e.g., “All 47 tests passed”).

voice.playback_diagnose

Diagnose playback system: device info, playback mode, env toggles, and optional test beep. Use when audio is silent or misconfigured.

Parameters:

• play_test_beep (boolean, optional) — play a short test chime (default: true).

voice.ambient_enable

Enable or disable the inner monologue (ambient) system at runtime. Does not require restart.

Parameters:

• enabled (boolean, required) — true to enable, false to disable.

voice.ambient_mute

Temporarily mute the inner monologue for a given number of minutes. The system remains enabled but suppresses output.

Parameters:

• minutes (integer, optional) — minutes to mute (default: 10, range 1 to 1440).