speech-model

ServiceNow 释放了 Apriel Voice 27B 预览版，一个基于 Gemma 4 骨干的开源语音-推理-工具调用模型，支持语音理解、多模态上下文和函数调用，可搭配外部语音合成器构建语音助手。

Dessa 推出一个仅基于语音数据训练的对话模型（Conversation Model），相关页面由 Block 管理，目前可见内容主要为网站隐私政策说明。

Block released Dessa, a full-duplex speech-to-speech voice model trained entirely on human speech that runs locally on consumer GPUs and produces expressive, human-like conversation with built-in emotion.

Decagon 推出其新型语音模型 Chord，由 Decagon Labs 团队开发，在 Modal 上完成训练与部署，具备低首帧音频延迟和高并发下的稳定表现。Chord 经过后训练以保持语速和语音自然度，并在所有部署中提升了问题解决率。

Wispr Flow introduces Canto, a speech model for real-time dictation that excels in noisy and challenging real-world conditions, achieving the lowest word error rate in evaluations against competitors.

DialectS2S is an end-to-end speech dialogue model for low-resource Chinese dialects, introducing a scalable data synthesis pipeline and a two-stage post-training strategy with self-aligned speech supervision. Experiments show improvements in dialect consistency, response quality, and intelligibility, with fully open-sourced models, data, and code.

Bland launches Speech v3, claiming it's the world's first Human Speech Engine and top model in Design Arena's Audio Realism benchmark, surpassing ElevenLabs, Grok, Cartesia, and OpenAI.

Microsoft is testing a new native real-time voice model, MAI Realtime, in early access on its MAI Playground. The full-duplex system supports multiple languages, low latency, and configurable turn-taking, positioning it as a competitor to OpenAI's GPT Live and Sesame.

A significant breakthrough in Levantine Arabic speech synthesis and English code-switching, achieving a 76% improvement using a single RTX 3060 in an evening.

This paper presents a case study using unsupervised articulatory probing to examine how self-supervised speech models encode phonetic features across Mandarin sub-dialects, finding that salient features like labiality remain stable while finer spectral distinctions show dialect-dependent variation.