Worktual Chatbot STT & TTS SDK

Chat widget with optional speech. The shop and its documents still come from My Account. The widget uses the same embed token as @worktual_limited/chatbot.

@worktual_limited/chatbot-stts

Overview

enableStt and enableTts are off unless you set them. Playback follows enableTts only. The answer text is revealed as a stream, and spoken audio starts with that text instead of after a second loading state.

enableStt

enableTts

Behavior

false

Text chat

true

false

Mic, then the transcript is the question. Text answer, no audio

false

true

Typed question, then the answer is spoken

true

Mic, then the transcript is the question, then the answer is spoken

Install

npm i @worktual_limited/chatbot-stts react react-dom

Use

import { WorktualChat } from "@worktual_limited/chatbot-stts";
import "@worktual_limited/chatbot-stts/ui/styles.css";

export function SupportChat({ token }: { token: string }) {
  return (
    <WorktualChat
      token={token}
      superDomainId="4204"
      subDomainId="1002"
      apiBase="https://unified-api.worktual.com"
      enableStt
      enableTts
    />
  );
}

Click the mic to record. Click it again to stop. The voice path is two calls to POST {apiBase}/chatbot-stts:

action: microphone (multipart audio) → { "text": "<user query>" } — SDK shows the user bubble.

action: voice_turn (JSON query + session) → gateway runs chat + TTS and returns answer text plus audio_base64.

Step

action

Body

Microphone

microphone

multipart: action + audio (16 kHz mono WAV)

Voice turn (mic + TTS)

voice_turn

JSON: session_id, query, tools → { answer, audio_base64, ... }. Text and audio start together

Typed chat

chat

JSON: session_id, query, tools (no audio)

Speak a typed answer

synthesize

JSON: text. Used only when TTS is on and the question was typed

End session

end_session

JSON: session_id

All calls send Authorization, Idempotency-Key, and subdomain.

In this document

Worktual Chatbot STT & TTS SDK

Overview

Install

Use

↑ Back to top