StepAudio 2.5
PulseAugur coverage of StepAudio 2.5 — every cluster mentioning StepAudio 2.5 across labs, papers, and developer communities, ranked by signal.
- 2026-05-11 research_milestone StepAudio 2.5 text-to-speech model achieved a top 3 global ranking, surpassing ElevenLabs v3. 来源
2 天有情绪数据
-
StepAudio 2.5 通过 RLHF 统一 ASR、TTS 和实时交互
一份新的技术报告介绍 StepAudio 2.5，这是一款统一的音频语言模型，旨在在自动语音识别 (ASR)、文本到语音合成 (TTS) 和实时语音交互方面表现出色。该模型通过针对任务的、来自人类反馈的强化学习 (RLHF) 来优化共享表示，从而实现这一点。这种方法允许一个单一的骨干模型被塑造成每个任务的独特操作模式，在标准基准测试中展示了最先进的性能。
-
StepAudio 2.5 TTS model ranks above ElevenLabs v3
StepAudio 2.5, a text-to-speech model from a Chinese AI lab, has reportedly surpassed ElevenLabs' v3 in performance, securing a top 3 ranking globally. The 24-month-old startup's model achieved this by outperforming Ele…
-
Jieyue AI's StepAudio 2.5 voice model ranks top in China, third globally
Jieyue AI has released its StepAudio 2.5 series of voice models, achieving a top global ranking in TTS performance. The StepAudio 2.5 TTS model specifically ranks third worldwide and first in China on the Artificial Ana…