<p>While working on voice agents and conversational systems this year, I kept running into the same problem again and again. Every time I needed a Text-to-Speech model, I had to dig through blog posts, GitHub repos, research papers, and product docs just to understand what actually existed and what was usable. Many lists were outdated, some mixed models from different years, and most didn’t explain where a model really shines or where it falls apart.
After doing this a few too many times, I decided to slow down and document what I was seeing. I put together a GitHub repo that focuses only on Text-to-Speech models released in 2025. It includes open-source models, research releases, and commercial APIs. For each one, I added the source link and simple notes on language support, strengths, and limitations, written from a builder’s point of view rather than a marketing one.
What stood out to me while compiling this was how diverse the space has become. Open-source dialogue models like Orpheus and Dia show how far natural speech and expressiveness have come. Research systems like MoE-TTS explore new ways of controlling voice using text descriptions. At the same time, commercial models from Qwen and Gemini still set the bar when it comes to latency and polish in production environments.
This repo isn’t meant to rank models or declare a “best” TTS system. It’s meant to be a clear, honest reference for anyone building voice agents, assistants, narration tools, or experimenting with speech generation in 2025. If you’re working in this space and want a quick overview without the noise, this might save you some time.
Sharing it here in case it’s useful. If you know a 2025 TTS model that’s missing, I’m happy to update it.</p>
<p><a href="https://github.com/pr0mila/Text-to-Speech-Models-Released-in-2025" target="_blank" rel="noopener nofollow">https://github.com/pr0mila/Text-to-Speech-Models-Released-in-2025</a></p>


Promila Ghosh

A curated GitHub repository lists Text-to-Speech models released specifically in 2025, including open-source, research, and commercial options. Each entry includes source links, language support details, and practical notes on strengths and limitations from a builder's perspective. The collection highlights the diversity of the TTS landscape, from expressive open-source dialogue models like Orpheus and Dia to research systems like MoE-TTS with text-based voice control, alongside production-ready commercial APIs from Qwen and Gemini.

A practical list of TTS models released in 2025