elevenlabs-ai

Original source / Raw SKILL.md

---
name: elevenlabs-ai
description: OpenClaw skill for ElevenLabs APIs: text-to-speech, speech-to-speech, realtime speech-to-text, voices/models, and dialogue workflows.
---

# ElevenLabs API Skill (Advanced)

## Purpose
Provide a production-oriented guide for using ElevenLabs APIs via direct HTTPS (no SDK requirement), with clear auth, safety, and workflow guidance.

## Best fit
- You need text-to-speech or speech-to-speech conversion.
- You want realtime speech-to-text with low latency.
- You prefer direct HTTP calls with predictable outputs.

## Not a fit
- You require a full SDK integration and helpers.
- You need full conversational agents beyond audio I/O.

## Quick orientation
- Read `references/elevenlabs-authentication.md` for API keys and single-use tokens.
- Read `references/elevenlabs-text-to-speech.md` for TTS endpoints and payloads.
- Read `references/elevenlabs-speech-to-speech.md` for voice conversion.
- Read `references/elevenlabs-speech-to-text-realtime.md` for realtime STT WebSocket.
- Read `references/elevenlabs-text-to-dialogue.md` for multi-voice dialogue output.
- Read `references/elevenlabs-voices-models.md` for voice IDs and model discovery.
- Read `references/elevenlabs-safety-and-privacy.md` for zero-retention and safety rules.

## Required inputs
- API key (xi-api-key) or a single-use token when needed.
- Voice IDs and model IDs for your target use case.
- Output format choice (audio codec/sample rate/bitrate).

## Expected output
- A clear workflow plan, endpoint checklist, and operational guardrails.

## Operational notes
- Keep a strict allowlist for downstream destinations of audio output.
- Cache voice IDs and model IDs server-side.
- Keep payloads small and retry with backoff on throttling.

## Security notes
- Never log API keys or tokens.
- Use single-use tokens for client-side access.


---

## Referenced Files

> The following files are referenced in this skill and included for context.

### references/elevenlabs-authentication.md

```markdown
# Authentication

## API keys
- ElevenLabs uses API keys for authentication.
- Send the key in the `xi-api-key` header on every request.
- Keys can be scoped and quota-limited; treat them as secrets.

## Single-use tokens
- Use single-use tokens when you must call from a client context.
- Tokens are time-limited and reduce key exposure risk.

```

### references/elevenlabs-text-to-speech.md

```markdown
# Text-to-Speech (TTS)

## Endpoint
- `POST /v1/text-to-speech/:voice_id`

## Core inputs
- `voice_id` (path)
- `text` (request body)
- `model_id` (optional)
- `output_format` (query param)

## Notes
- `enable_logging=false` enables zero-retention mode (enterprise only).

```

### references/elevenlabs-speech-to-speech.md

```markdown
# Speech-to-Speech (Voice Changer)

## Endpoint
- `POST /v1/speech-to-speech/:voice_id`

## Core inputs
- `voice_id` (path)
- audio input (file or stream per API)
- `output_format` (query param)

## Notes
- `enable_logging=false` enables zero-retention mode (enterprise only).

```

### references/elevenlabs-speech-to-text-realtime.md

```markdown
# Speech-to-Text (Realtime)

## WebSocket endpoint
- `wss://api.elevenlabs.io/v1/speech-to-text/realtime`

## Auth
- `xi-api-key` header, or
- `token` query param (single-use token for client-side sessions)

## Required params
- `model_id`

## Notes
- Supports partial and committed transcript events.
- Optional flags: timestamps, language detection, VAD settings.

```

### references/elevenlabs-text-to-dialogue.md

```markdown
# Text-to-Dialogue

## Endpoint
- `POST /v1/text-to-dialogue`

## Core inputs
- `inputs[]` list of `{ text, voice_id }` pairs
- `model_id` (optional)
- `output_format` (query param)

```

### references/elevenlabs-voices-models.md

```markdown
# Voices and Models

## Voices
- List voices: `GET /v1/voices`.
- Get a voice by ID: `GET /v1/voices/:voice_id`.

## Models
- Use `GET /v1/models` to list available model IDs.
- Prefer selecting models programmatically instead of hardcoding.

```

### references/elevenlabs-safety-and-privacy.md

```markdown
# Safety and Privacy

## Key risks
- Audio content can include sensitive data; treat outputs as PII.
- Avoid client-side API key exposure.

## Zero retention
- Some endpoints support `enable_logging=false` for zero retention (enterprise only).

## Operational safeguards
- Redact secrets in logs and monitoring.
- Use single-use tokens for browser-side calls.

```



---

## Skill Companion Files

> Additional files collected from the skill directory layout.

### _meta.json

```json
{
  "owner": "codedao12",
  "slug": "elevenlabs-ai",
  "displayName": "Elevenlabs AI",
  "latest": {
    "version": "1.0.0",
    "publishedAt": 1770631725619,
    "commit": "https://github.com/openclaw/skills/commit/5e908cab582652cde0737afd1120432c78fde1f5"
  },
  "history": []
}

```