# Bilibili Open-Sources Industrial-Grade Zero-Shot Voice Cloning Model IndexTTS-2.5

- Category: Flash (快讯)
- Published: 2026-08-12T11:58:00.000Z
- Source date: 2026-08-12
- HTML: https://www.88z.ai/en/news/flash-20260812-98cd4425ecbfc2b8
- Markdown: https://www.88z.ai/en/news/flash-20260812-98cd4425ecbfc2b8.md
- Other language: https://www.88z.ai/news/flash-20260812-98cd4425ecbfc2b8.md

Bilibili has open-sourced IndexTTS-2.5, an industrial-grade zero-shot voice cloning model. With only 0.8 billion parameters, it supports cross-language transfer across Chinese, English, Japanese, Spanish, and Arabic. The restructured inference architecture boosts speed by 2.28 times, achieving a real-time factor as low as 0.20. It also offers seamless speed adjustment from 0.5x to 2.0x and provides pronunciation intervention at three granularities: Pinyin, CMU phonemes, and Japanese Kana.

## At a glance

- IndexTTS-2.5 has only 0.8B parameters, supports cross-language transfer for five languages, and increases inference speed by 2.28 times.
- It supports seamless speed adjustment from 0.5x to 2.0x and offers three granularities for pronunciation intervention: Pinyin, CMU phonemes, and Japanese Kana.

## The update

Bilibili has officially open-sourced IndexTTS-2.5, an industrial-grade zero-shot voice cloning model. This model achieves robust functionality while remaining lightweight, with a parameter count of only 0.8 billion. It features strong multilingual capabilities, supporting cross-language voice transfer among Chinese, English, Japanese, Spanish, and Arabic. In terms of performance, IndexTTS-2.5 has restructured its inference architecture, increasing speed by 2.28 times and lowering the real-time factor to 0.20, which significantly improves processing efficiency. To cater to diverse application scenarios, the model supports seamless speed adjustment ranging from 0.5x to 2.0x. Furthermore, developers can finely intervene in pronunciation using three different granularities: Pinyin, CMU phonemes, and Japanese Kana, allowing for better control over the quality and accuracy of the output speech.

## Sources

- [IndexTTS-2.5](https://www.88z.ai/flash/cite/flash-20260812-98cd4425ecbfc2b8) (2026-08-12)

This article is an AI-assisted synthesis by 88Z AI based on the public sources below. Product claims and availability are attributed to their publishers. It does not represent independent testing or first-hand reporting.