User Guide

TTS API

Basic Usage

1. obtain keycloak access token.

2. make a POST request to TTS-API with desired text, speaker, sample rate and playback speed. the POST request returns a sessionID

3. the request is sent to the encoder (see Detailed System Description below for details), and generated audio is saved to minIO.

4. make a GET request to TTS-API with the sessionID to obtain a URL for the generated audio file.

API Documentation

documentation is available here 👈 Click

Text Input Formatting

TTS Text Normalization Rules

Reference : https://www.notion.so/atlaslabs/TTS-Text-Normalization-Rules-5b9c6b2364b74afba4429e6476fb54aa

Supported Character Set

• Korean characters, e.g. "안녕"

• English characters, upper-case and lower-case: "Hello" (transliterated if known, else spelled out e.g. A-P-P-L-E)

• Numbers: "1234"

• Symbols: space, punctuation: [ ! ? . , ], "-" for numbers (=="마이너스")

see the "Text Processing" section below for details on text processing