xAI 为 Grok Voice 推出 21 个多语言旗舰声音,原有五声音同步升级
21 New Flagship Grok Voices New multilingual voices for Grok Voice, plus improved naturalness for the original five. Jul 6, 2026
- 1xAI 发布 21 个 Grok Voice 旗舰新声音,与原有 5 个声音并行提供
- 2全部声音原生多语言,覆盖 Grok Voice 的 25+ 种语言,已上线 realtime Voice Agent API 和 Text to Speech API
- 3新的 Grok Voice Agent Builder 已在 xAI 控制台开放,用约一分钟音频即可克隆自己的声音
- 4原有五声音 Ara、Eve、Leo、Rex、Sal 经改进配方重训,官方称节奏、措辞与重音更自然
xAI 把语音能力从单一模型功能扩展成声音库、API 和搭建工具齐全的产品线,企业自建语音智能体的门槛在降低。
- 影响:多语言声音直接面向客服、广告、教育等场景,语音客服与语音内容可按语言统一选型;约一分钟音频即可克隆声音,自有品牌音色的获取成本下降;语音智能体搭建入口开放后,业务团队可自行尝试而无需先走开发。
- 参考:可考虑选一个多语言客服流程用 Voice Agent API 做小规模试点并记录话务成本对比;有品牌配音需求的团队值得用一分钟克隆音色测试相似度与稳定性,再决定是否调整现有配音预算。
- 依据:以上均为 xAI 官方发布口径,声音自然度与克隆效果暂无独立验证;25+ 种语言与约一分钟克隆音频的说法出自官方公告。
Today, we're releasing 21 new flagship voices for Grok Voice, joining the original five. All of them are multilingual and available now in the realtime Voice Agent API, the Text to Speech API, and the new Grok Voice Agent Builder. Meet the newcomers in the lineup:
Wellness
Support
Soft, empathetic, and soothing.
A voice for every use case
Each voice was cast for a specific job: support, characters, commentary, advertising, education. Speech tags like [pause] and <whisper> can be used to control the delivery. Hear five of them in their element:
Carina
Support
Thanks for your patience — I found the issue. [pause] Your account was still on the legacy plan, so I've moved you over and <emphasis>applied the credit to this month.</emphasis> <soft>You're all set.</soft> Anything else I can take care of?
Every voice is natively multilingual, supporting all of Grok Voice's 25+ languages.
Luna
闭上眼睛,静静聆听片刻。每一个声音都在讲述一个故事,每一个故事都值得被倾听。无论你身在世界的哪个角落,我都很高兴你在这里。
Improved naturalness for the original five
Ara, Eve, Leo, Rex, and Sal got an upgrade alongside the new voices. They've been retrained with an improved recipe, so pacing, phrasing, and emphasis are more natural.
Warm and friendly.
To build a custom voice agent with one of the new voices, use the Grok Voice Agent Builder in the xAI console. You can also clone your own voice from about a minute of audio.
来源:xAI:News · x.ai