ElevenLabs 在 ElevenAgents 推出合成语音检测并加入 Personal Agent Protocol
Defining how AI agents interact with your business
ElevenLabs 在 ElevenAgents 中发布合成语音检测功能，可在通话最初几秒分析来电者语音并判定其为真人还是 AI 生成，企业可据此设置规则，例如优先服务真人客户、限制 AI 来电者的交互范围，或在敏感操作前拦截克隆语音。该功能运行在实时音频流上，不依赖音频是否带有 ElevenLabs 水印。ElevenLabs 同时以设计伙伴身份加入由 Meta 和 Sierra 主导的 Personal Agent Protocol 工作组，参与制定个人智能体向企业表明身份、代表对象及可执行操作的规范。目前该检测功能面向由 Forward Deployed Engineering 团队支持的企业客户开放，本月晚些时候将作为 ElevenAgents 内的可配置选项向更广泛的企业客户提供。
推荐理由
AI 评分 51
ElevenLabs 在 ElevenAgents 中加入合成语音检测，并参与 Meta 与 Sierra 主导的 Personal Agent Protocol 标准制定，读者可了解企业如何区分来电者身份。
主题
正文
当前显示原文，中文全文译文尚未就绪。
Blog; Product
ListenListen to this article
On this page
Introduction; Detecting AI agents on live calls; Joining the Personal Agent Protocol; What's next
A growing share of the calls reaching businesses and governments aren't people. They're personal AI agents running errands for individuals, business agents confirming orders or checking claims, and occasionally a cloned voice posing as a customer.
Today, we’re releasing synthetic voice detection in ElevenAgents, so businesses can tell who’s on the line and respond accordingly. We’re also joining the Personal Agent Protocol working group as a design partner, to help define how AI agents and businesses should identify themselves and interact with each other.
Detecting AI agents on live calls
Synthetic voice detection analyzes the caller’s speech in the first few seconds of a call and returns a determination: human or AI generated. From there, your rules decide what happens.
A few examples of the rules you can set:
Serve verified human customers first. A human customer gets your most capable AI agent or a human representative, with everything that implies: moving to the front of a queue, negotiating a bill, applying a retention offer, or making a judgment call; Give AI callers direct, bounded exchanges. Machine callers go to an agent with its own instructions and limits. It authenticates first, then answers quickly and within scope. You decide how far it can go on negotiation, account changes, or exceptions. If the call does need a human, AI callers can sit lower in the queue than the people waiting; Stop impersonation at the start of the call. A synthetic voice requesting a sensitive action (such as a bank account change, a password reset, or access to another account) can be stopped before it gets information or takes action it shouldn’t
Synthetic voice detection runs inside ElevenAgents on the live audio stream, checking for traces a human listener wouldn’t notice but software can. All audio generated with ElevenLabs carries a watermark, but audio from other sources often doesn’t, so detection doesn’t depend on one being there.
Joining the Personal Agent Protocol
We’re working to design new standards for agents interacting with businesses, and we’re excited to contribute to the Personal Agent Protocol as part of the working group led by Meta and Sierra. The protocol is being developed to define how a personal agent identifies itself to a business, who it represents, and what it’s allowed to do on that person’s behalf.
Businesses use ElevenLabs to deploy agents across many text and voice channels: phone, website, in-app experiences, email, WhatsApp, SMS, and more. Our goal is to ensure all channels are covered, so customers can define how they interact with personal agents across every interface.
What's next
Today, synthetic voice detection is available to enterprise customers supported by our Forward Deployed Engineering team. Later this month, it will come to a broader group of enterprise customers as a configurable option inside ElevenAgents.
As standards like the Personal Agent Protocol take shape, we’ll add support for agents that declare themselves, and we’ll continue adding signals as the space matures.
To learn more, ask your account team or get in touch with us here.