Alibaba's Speech Large Model Tops Global Rankings, Earns National AI Triple Crown

On May 28, 2026, Artificial Analysis, a leading AI evaluation platform, published its latest Speech Arena rankings. Alibaba achieved a notable breakthrough with its voice large model Fun-Realtime-TTS-Preview, claiming fifth place globally and first in China with an Elo score of 1190.

1. Comprehensive Leadership: Leading Across Three Core Voice Tracks

In this evaluation, Alibaba's voice technology system showed strong overall performance, ranking first in China across three key voice AI tracks:

ASR (Automatic Speech Recognition): It achieved the top national ranking in accuracy and robustness for converting speech to text, showcasing Alibaba's ability to understand complex audio environments.

Chat (End-to-End Speech Understanding and Dialogue): It earned the top national ranking for real-time voice conversation fluency, logic, and response speed, reflecting Alibaba's industry-leading capabilities in intelligent assistant interaction.

TTS (Text-to-Speech): In this core competitive area, Fun-Realtime-TTS-Preview not only set new domestic records in naturalness, emotional expression, and rendering speed, but also established a global benchmark.

2. Technological Breakthrough: Real-Time Advancements with Fun-Realtime

The key player in this ranking—Fun-Realtime-TTS-Preview—marks a major breakthrough by Alibaba's voice team in real-time speech synthesis.

Previously, speech synthesis often struggled to balance high naturalness with fast response. Alibaba's model, however, successfully delivered speech output with near-human intonation at millisecond-level latency through an end-to-end deep architecture. This real-time capability is critical for time-sensitive applications like smart car interactions, digital human live streaming, real-time translation, and customer service.

3. Industry Insights: China's Voice Technology Advances Toward "Deep Intelligence"

As a trendsetter in the AI field, Artificial Analysis employs a highly rigorous scoring system that evaluates model performance on test sets and, more importantly, emphasizes user experience in real-world scenarios. Alibaba's "three championships" go beyond mere scores, conveying the following key insights:

Speech AI Enters the "Large Model Era": Earlier speech technologies largely depended on traditional statistical methods or small models. Alibaba's success demonstrates that embedding speech processing into a deep learning large model foundation can deliver a significant leap in perception quality.

"Chinese Speed" in Scenario Implementation: With Alibaba leading in both speech understanding and generation, future smart hardware and large model ecosystems in China will gain stronger global competitiveness in the core area of "voice interaction."

Illustration of Closed-Loop Capabilities: From recognition (ASR) to understanding (Chat) and then to synthesis (TTS), Alibaba has built a complete voice interaction pipeline, providing a solid foundation for developing seamless AI agents (Agents).

Through continuous bottom-up technical development and model iteration in the voice domain, AI in China is accelerating into deeper waters—moving from "being able to recognize" to "understanding human emotions and interaction logic" on a deeper level.