s2s-voice

A simple Speech-to-Speech library using Pollinations AI for Text-to-Speech (TTS) and Speech-to-Text (STT) in the browser. Works in React or plain JavaScript.

Install

npm install s2s-voice

Import the Library

import { playVoice, startSpeechToText, VOICES } from "s2s-voice";

Features

• Text-to-Speech (TTS): Convert text into human-like voice audio
• Speech-to-Text (STT): Capture user speech and convert it into text
• Works directly in React or any browser environment
• Lightweight, simple, and easy to integrate

Available Voices

You can select from the following voices when using playVoice:

export const VOICES = [
  "alloy", "serena", "aria", "luna", "kai",
  "milo", "nova", "sophia", "zen", "clio"
];

Default voice: "alloy"

Usage Examples

1. Text-to-Speech (TTS)

playVoice("Hello world", "alloy"); // default
playVoice("Hi there!", "serena");  // using another voice

• text → The text you want to speak
• selectedVoice → Optional, default "alloy"

2. Speech-to-Text (STT)

startSpeechToText((recognizedText) => {
  console.log("You said:", recognizedText);
});

• callback → Function that receives the recognized text

3. Example in React

import React from "react";
import { playVoice, startSpeechToText, VOICES } from "s2s-voice";

function App() {
  const handleSpeech = () => {
    startSpeechToText((text) => {
      console.log("Recognized:", text);
      playVoice(`You said: ${text}`, VOICES[1]); // "serena"
    });
  };

  return (
    <div>
      <button onClick={handleSpeech}>Start Speech-to-Speech</button>
    </div>
  );
}

export default App;

Quick Reference / Commands

| Task | Command | | ------------------------ | ------------------------------------------------------------ | | Install package | npm install s2s-voice | | Import functions | import { playVoice, startSpeechToText, VOICES } from "s2s-voice" | | Play text as speech | playVoice("Hello world", "voiceName") | | Start speech recognition | startSpeechToText(callback) | | List available voices | console.log(VOICES) |

Notes

• Works only in modern browsers
• Default voice: "alloy"
• Can be used directly in React or plain JS projects
• Speech recognition runs once per call; continuous recognition not supported