Alibaba's Qwen Audio 3.0 TTS Plus Tops the Speech Arena Text-to-Speech Leaderboard

In July 2026, Alibaba's Qwen Audio 3.0 TTS Plus took first place in Artificial Analysis's Speech Arena ranking. The text-to-speech model speaks in 16 languages and lets users set the speaking style with plain-language descriptions or tags such as [angry].

First place in Speech Arena

Qwen Audio 3.0 TTS Plus now leads Speech Arena, a public leaderboard of text-to-speech systems run by the analytics firm Artificial Analysis. The ranking compares voice models on synthesis quality, and Alibaba's new model came out ahead of its rivals overall.

• Developer: Alibaba, model Qwen Audio 3.0 TTS Plus
• First place in Artificial Analysis's Speech Arena ranking
• Support for 16 languages
• Speech style control: natural-language description or tags such as [angry]
• Generation speed: 16 characters per second, slower than Sonic 3.5 and Simba 3.2

How to give the voice an emotion

Qwen Audio 3.0 TTS Plus offers two ways to control speaking style: a natural-language description or special tags. For example, the [angry] tag makes the model sound irritated. This kind of control over intonation sets modern TTS models apart from the robotic speech synthesis of the previous generation, where the voice sounded flat and emotionless.

Support for 16 languages broadens the range of uses: the same model can voice content, power voice assistants and handle dubbing in different markets without switching engines.

Why the model is slower than its competitors

Qwen Audio 3.0 TTS Plus generates speech at 16 characters per second, noticeably slower than its direct rivals Sonic 3.5 and Simba 3.2. For offline tasks such as producing audiobooks or pre-recorded voiceovers, this is not critical: the file is rendered only once anyway. In real-time scenarios such as voice agents and live conversations, though, that speed means a noticeable delay between text input and sound.

The result is a familiar trade-off: Alibaba wins on synthesis quality, which is what Speech Arena measures, but loses on speed. For some tasks, natural-sounding voice matters more; for others, minimal latency does.

First place in

Speech Arena at a speed of just 16 characters per second: that is how the Artificial Analysis ranking describes the result for Qwen Audio 3.0 TTS Plus, according to The Decoder.

What this means

Qwen Audio 3.0 TTS Plus's lead shows that the TTS race has moved beyond simply "reading text intelligibly" to controlling emotion and style across many languages. But there is no universal winner yet: quality and speed remain a trade-off, and the right model depends on which matters more for the task.

Frequently asked questions

How many languages does Qwen Audio 3.0 TTS Plus support?

The model speaks text in 16 languages. It also lets users control speaking style and emotion through a natural-language description or special tags such as [angry].

Where does

Qwen Audio 3.0 TTS Plus fall short of its competitors?

Synthesis speed. The model generates 16 characters per second. According to Artificial Analysis, that is noticeably slower than competitors Sonic 3.5 and Simba 3.2, which is a problem for real-time scenarios.