Integrate the SDK › Implementation

Prerequisites and technical requirements

Before you write a line of code, the workspace has to hold what the integration will name, and your stack has to meet a few technical requirements. Access to the Speechineer service is either unauthenticated or authenticated, and the prerequisites and technical requirements differ accordingly — the items that apply to one of them only say so.

		Unauthenticated	Authenticated
Fit		public pages and development	anything behind a login
Who can use Speechineer in your form		anyone who can open the form	only users your backend issues a token to
API key	Mode	Unsigned	Signed
API key	Location	your frontend — the client passes it in its apiKey option	your backend — the token carries it in its api_key claim
Account	Location	your frontend — the client names it in its account option	your backend — the token carries it in its account_key claim
Signing key		not needed	needed — its private half signs every token

Prerequisites

A form to fill — either one configured in the workspace with a published version, as described in Configure the form › In the workspace, or one you configure in code, as described in Configure the form › In code.
An API key — in Unsigned mode for an unauthenticated client, in Signed mode for an authenticated one, as described in Authentication and usage › API keys › Auth mode.
A signing key pair (authenticated only) — its private half kept in your backend and its Key ID at hand, as described in Authentication and usage › Signing keys.
A stable key per end user — your own user id works well, as described in Authentication and usage › Accounts. Unauthenticated, you pass it as the account; authenticated, your backend signs it into the token.

Warning

The workspace Settings page on its Security tab, reached from Workspaces in the sidebar, with the Require signed account JWT switch highlighted — Workspaces › Settings › Security › Require signed account JWT — off for an unauthenticated form.

For an unauthenticated form, the workspace's Require signed account JWT setting has to be off, or every unsigned request is refused.

Technical requirements

Compatible frontend framework:
- React: 18 or newer.
- Angular: 17 or newer.
- JavaScript: any, with a bundler that handles ES modules.
Node.js 18 or newer — to install the package.
A secure origin — the page served over HTTPS (or localhost), since browsers only hand out the microphone there.
A backend that signs tokens (authenticated only) — one your users already log in to, since the token endpoint sits behind that login, with a library that signs PS256 tokens: jose in Node.js, PyJWT with the cryptography extra in Python, or your platform's equivalent.
The package for your framework:

bash

npm install @speechineer/react

bash

npm install @speechineer/angular

bash

npm install @speechineer/js