PlugWiki

speech-to-text Skill for Claude Code: Transcribe Audio with ElevenLabs Scribe

ElevenLabs skill that teaches your assistant to write code that turns audio and video into text with Scribe v2.

Installs
12.8K
on skills.sh
Stars
528
on GitHub
Files
6
besides SKILL.md
Updated
Oct 2026
last repo commit

When you ask for a transcript, subtitles or meeting notes from a recording, the skill tells the assistant how to call ElevenLabs Scribe v2 in Python, JavaScript or the ElevenLabs CLI. It covers word timestamps, speaker labels, multichannel audio, custom keyterms and live transcription.

It needs an ElevenLabs API key in ELEVENLABS_API_KEY and internet access. If no key is set, SKILL.md has the assistant run an onboarding command first. It bundles 6 reference files and no scripts.

Good for

  • Developers adding transcription to an app
  • Teams turning meeting or call recordings into text with speaker labels
  • Anyone who needs subtitles from audio or video files
  • Builders of live captioning or voice agents

Install speech-to-text

Claude Code

  1. 1

    Download the skill folder from https://github.com/elevenlabs/skills/tree/HEAD/speech-to-text.

  2. 2

    Put it at ~/.claude/skills/speech-to-text/ so that SKILL.md sits directly inside.

  3. 3

    Start a new session; the skill loads when its description matches the task.

The skill needs an ElevenLabs API key and internet access, and its setup step runs an npx command and installs SDK packages in your project, so this page lists no Claude app upload. The repo README also shows npx skills add elevenlabs/skills, which installs every ElevenLabs skill; you do not need it for this one.

Built from the package data we hold. Not yet tested by us.

What's in the skill

When it loads

Transcribe audio to text using ElevenLabs Scribe v2. Use when converting audio/video to text, generating subtitles, transcribing meetings, or processing spoken content.

Sections in SKILL.md

  1. Quick Start
  2. Python
  3. JavaScript
  4. CLI
  5. Models
  6. Transcription with Timestamps
  7. Speaker Diarization
  8. Multichannel Audio
  9. Keyterm Prompting
  10. Language Detection
  11. Supported Formats
  12. Response Format
  13. Error Handling
  14. Tracking Costs
  15. Real-Time Streaming
  16. Python (Server-Side)

Key facts

Author
ElevenLabs (owner of the elevenlabs/skills repo)
License
MIT (SKILL.md frontmatter)
Installs on skills.sh
12,774 as of 2026-11-02
GitHub stars
528 for the whole elevenlabs/skills repo, 2026-11-02
Last repo commit
2026-10-28
Files
7 files including SKILL.md (6 in references/); no scripts
Needs
ELEVENLABS_API_KEY and internet access (SKILL.md compatibility line)

What it includes

  • Python, JavaScript and CLI examples that transcribe one file with scribe_v2.

    Shows the shortest call in each language. The agent runs it when you ask for a transcript.

    Runs code · Quick start

  • Picks scribe_v2, scribe_v2_medical or one of three realtime models.

    Read · Models

  • Word-level start and end times, with word, spacing and audio_event types.

    Read · Timestamps

  • Labels who said what, with optional agent and customer roles for calls.

    Read · Speaker diarization

  • Handles one speaker per channel, as separate or merged transcripts.

    Read · Multichannel audio

  • Passes up to 100 terms the model might otherwise mishear.

    Read · Keyterm prompting

  • Detects the language or takes a language code as a hint.

    Read · Language detection

  • Live transcription with partial and committed transcripts and VAD commits.

    Covers server-side Python and client-side React, and when to commit.

    Network · Real-time streaming

  • Runs an npx onboarding command to sign in and store a key when none is set.

    Described as the first step before writing any code.

    Runs code · Onboarding setup step

From the official docs

Permissions

Medium risk

It can set up an API key and install packages in your project, and the code it writes sends audio to ElevenLabs. Approve each command and keep the key out of version control.

What it can do

  • Read your data
  • Create and change things
  • Delete things
  • Read-only mode

Watch out for

  • It may sign you in and store an API key

    If no ELEVENLABS_API_KEY is set, SKILL.md tells the agent to run an onboarding command from your app folder. Check where the key lands before you commit the project. github.com

  • It can install packages and tools

    The installation guide lists npm, pip, Homebrew, Scoop and a shell installer. Approve each command the agent proposes. github.com

  • Your audio goes to an online service

    SKILL.md says the skill requires internet access and an ElevenLabs API key. Do not send recordings you cannot share. github.com

Technical details
Sign-in and keysSKILL.md says the onboarding command signs you in and stores a key for the project; the reference guide also offers OAuth login that stores credentials in the OS keyring.
Scopes

  • Uses your ELEVENLABS_API_KEY

  • Agent may run npx, npm, pip or brew install commands from the reference guide

Questions

What does the speech-to-text skill do?

It teaches your assistant how to use ElevenLabs Scribe v2 to transcribe audio and video. The assistant writes code in Python, JavaScript or the CLI, with timestamps, speaker labels and live streaming options.

How do I install it?

Use the install box at the top of this page. The repo README also shows a command that installs every ElevenLabs skill at once, which you do not need for this one.

Does it need an API key?

Yes. SKILL.md says it needs an ElevenLabs API key in ELEVENLABS_API_KEY and internet access. If no key is set, it has the assistant run an onboarding command that signs you in and stores a key for the project. The key comes from your own ElevenLabs account, so calls count against that account's plan. SKILL.md states no prices or credit limits; its Tracking Costs section shows how to read the request-id response header to monitor usage. Our advice: check your plan before transcribing long files.

Is it safe?

It bundles no scripts, but the assistant it guides can run install commands and call an online API with your key. Read each command before you approve it and keep the key out of your repo.

What audio formats and sizes work?

SKILL.md lists MP3, WAV, M4A, FLAC, OGG, WebM, AAC, AIFF and Opus audio, plus common video formats. Limits are 5.0GB and 10 hours per file.

speech-to-text vs sound-effects or text-to-speech?

This one turns speech into text. The README lists text-to-speech and sound-effects as separate skills in the same repo for generating audio.

Information

updated Nov 2, 2026

Repository
elevenlabs/skills
Folder
speech-to-text

elevenlabs/skillsskills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 3
  1. SKILL.md · github.com
  2. README.md · github.com
  3. installation.md · github.com

Updated Nov 2, 2026.