Changelog

April 25, 2025

SDK: New Typescript SDK

We’ve published a new Typescript SDK. Try it out:

$npm install @speechify/api

Together with the release, the old @speechify/api-sdk package is now deprecated.

Migrating from @speechify/api-sdk

The JWT-based authentication endpoint has been deprecated. This means that the auto-refresh token feature has also been removed in the new SDK.

  • How to call individual endpoints is available in the API Explorer. Switch to the Typescript SDK tab to see the snippet.
  • The full implementation of the new SDK is available in the GitHub repository.

Quick Start Guide

1

Install the SDK

Install the package using npm:

$npm install @speechify/api
2

Initialize the Client

Import the SDK and create a new client instance:

1import { SpeechifyClient } from "@speechify/api";
2
3// API Key can also be set as an environment variable SPEECHIFY_API_KEY
4const client = new SpeechifyClient({ apiKey: "your-api-key" });
3

Fetch Available Voices

Get a list of all available voices:

1const voices = await client.tts.voices.list();
4

Generate Speech

Create a new speech synthesis request using a voice:

1const result = await client.tts.audio.speech({
2  voiceId: voices[0].id,
3  input: "Hello, world!",
4});
5

Save the Audio

This is now an additional required step since the SDK only returns the raw audio data i.e. a string representing the base64 encoded bytes. Convert the audio data and save it to a file:

1const audioData = result.audioData;
2const audioBuffer = Buffer.from(audioData, "base64");
3
4// Write the binary data to a file
5fs.writeFileSync("speech.mp3", audioBuffer);

API: Deprecated POST /v1/auth/token endpoint

The POST /v1/auth/token endpoint is now deprecated. We will only support API Key authentication moving forward.

March 24, 2025

API: Text Normalization

We’ve added a new option to the API to normalize the text. If enabled, it will transform numbers, dates, etc. into words. For example, “55” is normalized into “fifty five”. This can increase latency due to additional processing required for text normalization.

March 21, 2025

API: Loudness Normalization Update

Previously, the loudness normalization was enabled by default. Now it’s disabled by default as we believe, together with our internal model improvements, it will improve the overall quality of the audio.

SDK: Python

We’ve released a new Python SDK. Checkout the repository for more information.