CHANGELOG

What's New

Latest updates, features, and improvements

February 2026

6 New TTS Engines
Feb 10
New
  • Added CosyVoice2 - zero-shot multilingual voice cloning (Apache 2.0)
  • Added GPT-SoVITS - highest quality few-shot voice cloning (MIT)
  • Added Dia - dialogue generation with nonverbal sounds (Apache 2.0)
  • Added Qwen3-TTS - 10-language TTS with 3-second cloning (Apache 2.0)
  • Added Zonos - expressive TTS with emotion control (Apache 2.0)
  • Added Pocket TTS - ultra-lightweight CPU voice cloning (CC-BY-4.0)
  • Platform now supports 17 TTS engines with all commercially-safe licenses

January 2026

Voice Cloning Improvements
Jan 26
New Improvement
  • Added GPU OOM retry/failover system for more reliable processing
  • Improved voice cloning quality with better sample preprocessing
  • Fixed demo generation issues with null user handling
  • Added CSRF_TRUSTED_ORIGINS for Django 4.0+ compatibility
GPU Infrastructure Upgrade
Jan 20
New
  • Deployed 4x Tesla P40 GPUs (96GB total VRAM)
  • Added dynamic GPU allocation with Redis-based memory locking
  • Implemented tiered queue system (tts-fast, tts-standard, tts-premium)
  • Voice sample generation now runs on GPU for faster processing
Demo Mode Enhancements
Jan 15
Improvement
  • Demo mode now hides credit estimates for cleaner UX
  • Filtered premium voices from demo to show accessible options
  • Improved voice cloning landing page experience

December 2025

Platform Launch
Dec 15
Launch
  • Initial public launch of TextToSpeechAI
  • Support for 11 TTS engines: Piper, VITS, Bark, StyleTTS2, F5-TTS, Parler, OpenVoice, Tortoise, Kokoro, MeloTTS, Chatterbox
  • 350+ AI voices across 50+ languages
  • Voice cloning with F5-TTS, StyleTTS2, and OpenVoice
  • Credits-based billing system
  • REST API for developers
Legend
New New feature
Improvement Enhancement
Fix Bug fix
Launch Major release

Stay Updated

Create an account to receive updates about new features and improvements.

Create Free Account