Anthropic 或自研语音模型,Claude 现隐藏“Voice”入口
TestingCatalog 发现 Claude 网页导航中出现了一个单独的隐藏“Voice”条目,并带有“Preview”标签;该标签指向 Anthropic 的内部预览环境,而非公开发布,目前也没有关于该界面功能或它与现有语音模式差异的细节。该发现紧随 Anthropic 近期推出的一项可选设置:允许用户分享语音录音和对话数据用于模型训练,该设置与文本对话的授权机制相互独立,且默认关闭。报道指出,Anthropic 尚未通过 API 公开发布自有的专用文本转语音模型,也未推出原生端到端实时音频模型,因此目前无法确定它是在开发实时语音模型、重做界面,还是测试其他内容。
报道指出,Anthropic 至今未通过 API 发布专用文本转语音模型,也没有原生实时音频模型,而该隐藏入口的具体用途尚不明确。
Anthropic appears to be experimenting with a new voice interface for Claude. We spotted a separate hidden “Voice” item in the web navigation labeled “Preview.” This label refers to Anthropic’s internal preview environment rather than a public release. No details have surfaced about what the interface would offer or how it might differ from Claude’s existing voice mode.
The discovery follows Anthropic’s recent introduction of an optional setting that allows users to share voice recordings and conversation data for AI model training. It is separate from the existing consent mechanism for text conversations and is disabled by default. Collecting this data could indicate that Anthropic is preparing to develop its own speech models, although it could also support work on existing voice capabilities.
Anthropic starts prompting users to voluntarily share their voice conversations for training purposes. > “Allow us to use your voice data to improve our AI models” > “Audio recordings and voice chat data help improve how Anthropic AI models understand and respond to speech.… pic.twitter.com/KUxB8OB88b— 🚨 AI News | TestingCatalog (@testingcatalog) October 4, 2026
Claude already supports voice conversations across mobile, desktop, and web through a system that processes speech using its language models and generates spoken responses. Anthropic previously held discussions with ElevenLabs about potential voice partnerships, but the underlying provider has not been officially confirmed. The company has also neither publicly released its own dedicated text-to-speech models through its API nor introduced a native end-to-end real-time audio model.
A dedicated Voice interface could eventually become central to "vibe coding" workflows, potentially enabling more natural conversations with Claude while it performs tasks. However, it remains unclear whether Anthropic is developing a real-time voice model, redesigning the interface, or testing something entirely different. There is also no indication of when this experiment could become publicly available.
Sources and related context
- Claude Help Center explains existing voice mode: Supports: Anthropic documents spoken conversations on mobile, desktop, and web using the same Claude models available in text chat.
- BleepingComputer reports Claude's separate voice-data training setting: Supports: The report documents the optional voice-recording and voice-chat training setting, its separation from text-training consent, and its disabled-by-default state in the observed interface.