Quick Ref / Command

Marketing Kit

/ak:elevenlabs

ElevenLabs audio generation

Guides speech generation, authorized instant voice cloning, and sound-effect creation with ElevenLabs models and scripts; music and voice-agent topics are reference-only.

01

Pick action

02

Select model

03

Load guide

04

Tune voice

Rule 01

Voice selection matters more than model choice, then settings refine the result.

Rule 02

Latency-sensitive speech drafts can consider Flash v2.5 with streaming, but no live voice-agent route is implied.

Rule 03

Repeated audio should be cached instead of regenerated.

Rule 04

Pronunciation and emotion need deliberate controls, not default output acceptance.

Execution Map

Core lanes from input to output, aligned with the command's real execution path.

01

Start

Input, scope, route

  1. 1 Pick action Read speak, clone, or sfx from the request and map it to the right ElevenLabs capability.
  2. 2 Select model Choose Multilingual v2 for quality, Flash v2.5 for real-time latency, Turbo v2.5 for balance, or Eleven v3 for emotional range.
02

Work

Agent / skill execution

  1. 3 Load guide Use the relevant direct-action reference for text-to-speech, instant voice cloning, or sound effects; broader music and agent material is knowledge-only.
  2. 4 Tune voice Set voice, stability, similarity, speed, pronunciation, pauses, and emotional controls for the medium.
03

Verify

Gate, review, validation

  1. 5 Generate audio Run the matching SDK or script for TTS, voice management, cloning, or sound-effect generation.
  2. 6 Optimize cost Cache repeated audio and prefer Flash models when latency or credit usage matters.
04

Close

Report, handoff, artifact

  1. 7 Prepare handoff Return model choice, voice settings, scripts used, generated assets, and any streaming or phone-integration notes.

Special Operations

Text-to-Speech

Convert copy into lifelike audio with voice settings, SSML-style pauses, pronunciation, and streaming choices.

Instant voice cloning

Create authorized instant clones from clean samples; professional-cloning training is outside the helper route.

Reference-only topics

Use music, voice design, and conversational-agent material as planning knowledge, not as dedicated invocation routes.

Sample Prompt

Concrete invocations for each flag, subcommand, mode, or route available in the live workflow.

Generate speech Recommended
/ak:elevenlabs speak "Welcome to the product tour"
Use when:
You need lifelike spoken audio from text.
Expected:
Chooses a TTS model, voice settings, and generation path.
Clone voice
/ak:elevenlabs clone voice-sample.wav
Use when:
Authorized recordings should become an instant clone in the provider account.
Expected:
Applies instant voice-cloning requirements, consent checks, and recording best practices.
Sound effect
/ak:elevenlabs sfx "short success chime for checkout"
Use when:
A short generated sound effect is needed.
Expected:
Uses the sound-effects guidance, duration limits, and prompt structure.

Handled Scope

  • Text-to-speech
  • Authorized instant voice cloning
  • Sound effects
  • Consent and rights review
  • Provider cost boundaries

Next