Contribute — speech.dev

Request new voice models, labs, skills, templates, or facts-layer corrections via structured GitHub issues.

How it works

  1. Step 1

    Pick a request type

    Use the form that matches your change. Include primary source URLs — pricing pages, API docs, changelogs.

  2. Step 2

    Maintainer triage

    We check ids and sources. Incomplete requests get a follow-up; complete ones get the agent label.

  3. Step 3

    Agent → draft PR

    The content agent writes JSON as draft, runs validate + build, and opens a PR for human review.

Request types

Each button opens a structured GitHub issue. You prepare ids and sources; the agent structures JSON. All submissions stay draft until a maintainer promotes them.

New model lab
Add a publisher not yet listed under /data/model-labs/.
Request new lab
New voice model
Add a variant (separate id) under an existing or new lab.
Request new model
Update voice model
Pricing, languages, status, docs, or other facts changed.
Update a model
Update model lab
Lab name, homepage URL, or status changed.
Update a lab
Skill
Installable voice skill for AI coding tools.
Add or update skill
Template
Open-source voice agent starter repository.
Add or update template
Taxonomy tag
Platform, telephony, or other controlled-list entry.
Change taxonomy

Before you submit

  • Facts layer only — not operational notes from production unless invited
  • One primary source URL per factual claim
  • Search existing entries to avoid duplicates
Browse all templatesRead contribution guide

On model pages

Every voice model detail page includes a Suggest an update panel with prefilled issue links for that entry and lab.

Data verification explains how facts and operational notes differ.