Summary

IndexTTS 2.5 is a zero-shot neural text-to-speech foundation model for expressive multilingual voice synthesis. It builds on a transformer-based Text-to-Semantic module and a non-autoregressive Semantic-to-Mel module to reproduce voice characteristics and emotion.

Key Features

Zero-shot neural text-to-speech.
Transformer Text-to-Semantic module.
Non-autoregressive Semantic-to-Mel module.
Emotion and voice characteristic replication.
25 Hz semantic codec compression.
Zipformer S2M architecture.
Chinese, English, Japanese, Spanish, and Arabic.
Cross-lingual emotion transfer.

About IndexTTS 2.5


The update compresses the semantic codec from 50 Hz to 25 Hz, reducing sequence length and inference cost, and replaces the earlier S2M backbone with a more efficient Zipformer architecture. New cross-lingual strategies use boundary-aware alignment, token-level concatenation, and instruction-guided generation.


IndexTTS 2.5 supports Chinese, English, Japanese, Spanish, and Arabic, including emotion transfer without target-language emotional training data. It is useful for narration, localization, character voices, accessibility, and research into zero-shot emotional TTS.

Get more likes & reach the top of search results by adding this button on your site!

Embed button preview - Light theme
Embed button preview - Dark theme
TurboType Banner

Subscribe to the AI Search Newsletter

Get top updates in AI to your inbox every weekend. It's free!