Getting started
Introduction
Before you write a line of code, it helps to know what Speechineer does and does not do, how an integration is organised, and the vocabulary the rest of this manual uses — workspaces, forms, API keys, signing keys, accounts and sessions. This page introduces them once; every later page assumes you have read it.
How to use Speechineer
- Configure the form. Decide what to extract — fields with stable ids and the prompts extraction follows — either in your workspace (publish a version) or as a form object in your code. Configure the form.
- Authentication and usage. Create an API key in the right mode — Unsigned for public pages and development, Signed (with a signing key) for anything behind a login — decide what an account is for your users, and set the caps. Authentication and usage.
- Integrate the SDK. Create a client once with those credentials, one session per form, and one button. The user talks; values stream into
valuesandonFieldValue, keyed by your field ids. Integrate the SDK.
Note
Once sessions run, every unit they consume is attributed to the workspace, the API key, the account, the form and the session it ran in, and stops at whatever limits you set. The workspace's Usage page shows this as it happens.
Vocabulary
| Term | What it is | Section |
|---|---|---|
| Workspace | One piece of your software — an app, a service, a component. It holds that software's forms, keys and accounts, and carries its own limits, so usage is reported per software. Dev, staging and production share one workspace, told apart by their API keys. | Getting started |
| Form | What to extract: fields with stable ids, the prompts extraction follows, and the model configuration — configured in the workspace, or defined in your code. | Configure the form |
| API key | How a client identifies the workspace. Each key has a mode — Unsigned (the key rides in your frontend: development and public pages) or Signed (only tokens your backend signs are accepted: anything behind a login). | Authentication and usage |
| Signing key | The key pair behind Signed mode. The public half lives in your workspace; the private half stays in your backend and signs a short-lived token per user. Rotates with a grace period. | |
| Account | The end user a session works for — a stable key you choose, one per user or tenant. Every unit consumed is attributed to it, and limits can be set on it. | |
| Session | One run of filling one form: it listens, recognizes values, and reports one observable state your UI renders from. | Integrate the SDK |
What Speechineer does not do
- It never touches your DOM. There is no widget to embed and no styling to adopt — you render everything, including the talk button.
- It never owns your data model. Values arrive as plain key–value pairs; your validation and your submit decide what happens to them.
- It never requires a form rewrite. The one contract is that field ids match — everything else about your form is yours.
Ready to see it in a diff? Quickstart shows the before → after for React, Angular and plain JavaScript.