# Qwen3-TTS Technical Report

Source: https://www.youtube.com/watch?v=A_WX7j1TlXk
Recap page: https://rapidrecap.app/video/A_WX7j1TlXk
Generated: 2026-01-23T16:06:54.138+00:00

---
## Quick Overview

The Qwen3-TTS technical report reveals that the 12-Hertz model significantly outperforms the 25-Hertz model, achieving a 66% reduction in word error rate compared to CozyVoice 3, due to its ability to process entire sentences contextually rather than chunking audio, which leads to better prosody, tone, and overall human-like speech quality, despite the larger computational demands of the multi-stage architecture.

**Key Points:**
- The 12-Hertz model achieved a 66% reduction in word error rate compared to the CozyVoice 3 benchmark.
- The 12-Hertz model uses a multi-stage process involving a flow-matching-based diffusion transformer, unlike chunk-based systems.
- The 12-Hertz model processes entire sentences contextually, resulting in superior prosody, tone, and accent matching.
- The 25-Hertz model, while fast, performed worse than the 12-Hertz model, especially in challenging tasks like German translation.
- The 12-Hertz model's audio quality was rated subjectively as 'beautiful' and 'perfect', integrating well with semantic understanding.
- Qwen3-TTS utilizes two tokenizers, one capturing semantics and another capturing acoustics, for efficient processing.
- The model's architecture forces it to plan the audio generation ahead of time, avoiding stuttering or immediate errors common in reactive models.

**Context:** The video discusses the technical advancements presented in the Qwen3-TTS technical report, released on Thursday, January 26, 2022. The report focuses on improving Text-to-Speech (TTS) models, specifically comparing the performance of two new models, a 12-Hertz version and a 25-Hertz version, against existing state-of-the-art models like CozyVoice 3 and models based on chunking.

## Detailed Analysis

The Qwen3-TTS technical report details improvements in synthetic speech generation, primarily comparing the 12-Hertz and 25-Hertz models. The 12-Hertz model, despite being slower, demonstrated superior performance, achieving a 66% reduction in word error rate compared to CozyVoice 3 on a specific benchmark. This improvement stems from its architectural difference: it uses a flow-matching-based diffusion transformer that processes audio contextually over entire sentences, rather than processing in small chunks like older models. This context-aware processing allows it to maintain natural prosody, tone, and accurately mimic speaker identity, leading to higher perceived quality. The 25-Hertz model, while faster (generating output in 190 milliseconds), showed lower semantic fidelity, especially on difficult tasks like German translation, where it struggled with rhythm and intonation compared to English. The architectural advantage lies in its multi-stage approach: one layer captures semantics (meaning), and subsequent layers capture acoustics (tone, noise, breathing, accent). Furthermore, it uses a causal mechanism, planning ahead rather than reacting immediately, which avoids common TTS artifacts like stuttering or garbled speech. The paper suggests that for complex emotional tasks, this look-ahead planning is crucial. The primary takeaway is that the Qwen3-TTS architecture marks a generational leap over previous methods that relied on chunking or immediate reaction, pushing audio quality closer to human perception.

### Qwen3-TTS Model Comparison

- 12-Hertz model achieved 66% lower WER than CozyVoice 3
- 25-Hertz model is faster but has lower semantic fidelity

### Architectural Differences

- 12-Hertz uses flow matching/diffusion transformer for contextual processing
- 25-Hertz uses a different architecture that struggles with long sequences

### Key Features

- Multi-layered processing (semantics then acoustics)
- Causal mechanism allows for future planning
- Excellent handling of prosody and tone across 10 languages

### Performance Metrics

- 12-Hertz model achieved 22% word error rate on a specific test
- outperformed older models like 11 Labs and MiniMax

### Implications

- The future of TTS lies in context-aware, multi-stage processing that minimizes latency while maximizing human-like naturalness.

![Screenshot at 00:00: Opening screen displaying the podcast/speaker graphic and the call to 'Become a Member Today!'](https://ss.rapidrecap.app/screens/A_WX7j1TlXk/00-00-00.jpg)
![Screenshot at 02:03: Visual representation of the two tokenizers being discussed: semantic and acoustic.](https://ss.rapidrecap.app/screens/A_WX7j1TlXk/00-02-03.jpg)
![Screenshot at 05:55: Visual comparison showing the fast generation speed of the 25-Hertz model versus the need for longer context.](https://ss.rapidrecap.app/screens/A_WX7j1TlXk/00-05-55.jpg)
![Screenshot at 08:37: Visual representation of DPO \(Direct Preference Optimization\) being used to align the model with human preference.](https://ss.rapidrecap.app/screens/A_WX7j1TlXk/00-08-37.jpg)
![Screenshot at 11:16: A representation of the 12-Hertz model's superior performance in handling complex instructions like speaking in a sad whisper.](https://ss.rapidrecap.app/screens/A_WX7j1TlXk/00-11-16.jpg)