openai-fm-tts

A TypeScript module for generating Text-to-Speech (TTS) audio using an unofficial, publicly accessible /generate route on https://www.openai.fm/. This package allows customization of voices and vibes, with support for saving audio in WAV or MP3 formats.

Note: This module uses an undocumented API endpoint, which may not be officially supported or guaranteed to remain available.

Features

• Generate TTS audio with customizable voices (e.g., Alloy, Nova, Shimmer).
• Apply unique "vibes" (e.g., Pirate, Cyberpunk Hacker, Fairy Tale Narrator) to alter tone, emotion, and style.
• Save audio output in WAV or MP3 formats using FFmpeg for MP3 conversion.
• TypeScript support with comprehensive type definitions.
• Modular design compatible with Bun and Node.js environments.

Installation

npm install openai-fm-tts

Ensure you have the following dependencies installed:

• TypeScript (>=5.0.0)
• FFmpeg (required for MP3 output)

Usage

Generating TTS and Saving Audio

import { generateTTS, saveAudio, Voice, VibeList, vibes } from "openai-fm-tts";

const vibePerson = vibes[VibeList.SciFiCommander];

async function example() {
  try {
    const { audio, filename } = await generateTTS({
      input: "Welcome to the future!",
      voice: Voice.Echo,
      vibe: vibePerson.name,
    });
    const outputPath = await saveAudio({ audio, filename }, "output", "mp3");
    console.log(`Audio saved to: ${outputPath}`);
  } catch (error) {
    console.error("Error:", error);
  }
}

example();

API

generateTTS(request: GenerateRequest): Promise<GenerateResponse>

Generates TTS audio by sending a request to the /generate endpoint.

• Parameters:
  • input: The text to convert to speech.
  • prompt: Optional custom prompt for additional context.
  • voice: Optional voice selection (defaults to Voice.Nova).
  • vibe: Optional vibe to apply (e.g., VibeList.Pirate).
• Returns: A promise resolving to { audio: ArrayBuffer | Uint8Array, filename: string }.

saveAudio(response: GenerateResponse, outputPath: string, format: "wav" | "mp3"): Promise<string>

Saves the generated audio to a file.

• Parameters:
  • response: The response from generateTTS.
  • outputPath: The desired file path (without extension).
  • format: The output format (wav or mp3).
• Returns: A promise resolving to the saved file path.

Voices and Vibes

Voices

Available voices include: Alloy, Ash, Ballad, Coral, Echo, Fable, Onyx, Nova, Sage, Shimmer, Verse.

Vibes

Vibes add personality to the audio, such as:

• Medieval Knight: Noble and heroic.
• Pirate: Boisterous with seafaring charm.
• Cyberpunk Hacker: Edgy and tech-savvy.
• Fairy Tale Narrator: Whimsical and magical.
• See VibeList enum for the full list.

Requirements

• Node.js or Bun runtime.
• FFmpeg installed for MP3 conversion.
• TypeScript for development (optional but recommended).
• tsup for building the project (required for the build script).

Building

npm run build

This compiles the TypeScript source to JavaScript in the dist/ directory and generates type declarations in the dist/index.d.ts file.

Disclaimer

This package relies on an unofficial API endpoint (/generate) from https://www.openai.fm/, which is publicly accessible but not documented. Usage of this endpoint may be subject to change or discontinuation without notice.

License

MIT