“Voice User Interface (VUI) Designer” is an increasingly prominent job title in the tech world. A VUI designer typically writes the conversation and designs the flow between a VUI — an invisible interface that communicates through the medium of voice, like Alexa or Google Home — and the end user. The discipline of VUI Design, a subset of conversation design is, as leading VUI designer, Cathy Pearl explains on her website, “about teaching computers to communicate like humans, and not the other way around.”

“语音用户界面(VUI)设计器” 在科技界越来越重要。 VUI设计人员通常会编写对话并设计VUI(即通过语音介质(例如Alexa或Google Home)进行通信的不可见界面)与最终用户之间的流程。 正如VUI首席设计师Cathy Pearl在她的网站上解释的那样，VUI设计这一学科是对话设计的一个子集，“关于教计算机像人一样交流，而不是相反。”

As the world has found increasing value in VUIs, the tech industry in turn has invested more in producing voice-forward products.

随着世界范围内VUI的价值不断增长，技术行业反过来又在生产语音转发产品上投入了更多的资金。

Designers with a background in only screen-based products, also known as Graphic User Interfaces (GUIs), will find that designing for VUIs is quite contrast. While there is plenty of overlap in the necessary skills to design for both formats (having empathy for users, etc.), the two interface types have a very different communication landscape.

仅在基于屏幕的产品(也称为图形用户界面(GUI))中有背景的设计人员会发现，针对VUI进行设计是完全相反的。 虽然两种格式的必要技能有很多重叠(对用户有同感等)，但两种界面类型的通信方式却大不相同。

通过GUI与VUI进行交互 (Interaction over GUIs versus VUIs)

Interaction between a system and a user looks (or sounds, rather) very different when it takes place over a VUI versus over a traditional GUI.

当系统和用户通过VUI与传统GUI进行交互时，它们之间的交互(或听起来)相当不同。

A GUI communicates to a user via screens comprised of visual design and text, whereas a VUI communicates via audio. Consequently, a user’s senses (visual, haptic, audio, etc.) that these two interfaces signal to in order to help send a message, are different.

GUI通过包含视觉设计和文本的屏幕与用户通信，而GUI VUI通过音频进行通信。 因此，这两个接口发信号通知以帮助发送消息的用户感觉(视觉，触觉，音频等)是不同的。

When it comes to absorbing and processing information laid out on an interface, a user does so very differently on a GUI versus on a VUI. On a GUI, a user moves at their own pace; they ponder, review, and skip ahead at their own leisure. On a VUI on the other hand, a user is less in control when it comes to how they receive information. James P. Giangola and Jennifer Balogh explain in their book, Voice User Interface Design, that a VUI “present(s) information serially and non-persistently” which makes a higher cognitive demand on users than does a GUI by “further challenge(ing) human memory and attention.” Moreover, when it comes to audio, listeners’ “ears can only handle one fact or idea per sentence” according to NPR, less than our eyes can handles when reading.

当涉及到吸收和处理界面上布置的信息时，用户在GUI和VUI上的操作截然不同。 在GUI上，用户按照自己的步调移动 ； 他们思考，回顾并在闲暇时跳过。 另一方面，在VUI上，用户在接收信息方面的控制较少。 James P. Giangola和Jennifer Balogh在他们的《 语音用户界面设计 》一书中解释说，VUI“连续且非持久地呈现信息”，与GUI相比，通过“ 进一步挑战 ，VUI对用户的认知要求更高”。 )人类的记忆力和注意力 。” 此外， 在音频方面，听众的“耳朵每个句子只能处理一个事实或想法” 根据NPR ，比阅读时我们的眼睛所能承受的少。

To send input back to a GUI, a user interacts with what is laid out on the screen, often responding back with taps and other motions. When a user wants to respond to a VUI, they are not bounded by buttons and character limits like they are on a GUI. They speak their responses back, which allows them much more freedom.

为了将输入发送回GUI，用户需要与屏幕上显示的内容进行交互，并经常用敲击和其他动作来响应。 当用户想要响应VUI时，它们不受GUI上的按钮和字符限制的限制。 他们说出自己的回答 ，这给了他们更多的自由。

With a new communication landscape comes new Best Practices. I’ve compiled a list of tips for GUI designers who want to design for VUIs that will engender a positive user experience.

新的沟通方式带来了新的最佳实践。 我为希望为VUI设计将带来积极的用户体验的GUI设计人员汇总了一些提示。

1. CTAs应该做好准备 (1. CTAs should set the stage)

Write calls to action that specify what exactly your VUI can do.

编写号召性用语，以指定您的VUI可以做什么。

When interacting with a voice assistant, a user is unbounded in the range of responses with which they could reply. Just about the worst prompt you can program your assistant to begin with is the open-ended “How can I help you?” question.

与语音助手交互时，用户在可以回复的响应范围内不受限制。 您可以为助手编程的最糟糕的提示是：不限成员名额“我如何帮助您？” 题。

As an example, let’s say you’re designing “Substitution Sidekick,” a cooking and baking skill that can offer a user substitute ingredients if they are out of the original ingredient that a recipe calls for. Your assistant opens with “What can I help you with today?” and your user responds with “Where can I buy ghee?” Your skill isn’t here to inform users about the whereabouts of certain grocery items, and so you never programmed it to respond to these types of questions. This “edge case”, a rare situation that threatens to break the user experience, could have been avoided if you had begun the conversation with a more targeted call to action, like:

举例来说，假设您正在设计“ Substitution Sidekick”，这是一种烹饪和烘焙技能，如果用户使用的配方不需要原始成分，则可以为用户提供替代成分。 您的助手将打开“今天有什么可以帮助您的？” 然后您的用户回答“我在哪里可以买酥油？” 您的技巧不是在通知用户某些杂货的下落，因此您从来没有编程它来回答这些类型的问题。 如果您以更具针对性的号召性用语来开始对话，则可以避免这种“边缘情况”，这种罕见情况有可能破坏用户体验，例如：

“Tell me the ingredient you’d like to substitute today.”

“告诉我您今天想替代的原料。”

2.设置成功的用户 (2. Set the User up for Success)

Be strategic with how you frame prompts.

制定提示框架时要有策略性。

Frame your VUI’s lines in a way that will elicit the response your system needs to hear to be effective. Humans and computers communicate in different ways. Conversation designers are here to bridge the gap between the two and ensure we create experiences that don’t force humans to adapt to computer-speak.

对VUI的线路进行构架，以引起系统需要听到的有效响应。 人与计算机的通信方式不同。 对话设计师在这里弥合两者之间的鸿沟，并确保我们创造的体验不会强迫人类适应计算机语言。

Say your user tells “Substitution Sidekick” they are out of milk. Your VUI probes them further and says, “Ok, you’re looking for a substitute for milk. Is this for a cooked dish, a baked good, or a fresh meal?” and your user responds with “yes.” Which one were they responding “yes” to?

假设您的用户告诉“ Substitution Sidekick”他们没有牛奶。 您的VUI会进一步探查它们，并说：“好吧，您正在寻找牛奶的替代品。 这是煮熟的食物，烤制的食物还是新鲜的饭菜？” 并且您的用户回答“是”。 他们对哪个人回答“是”？

Be explicit, but still conversational; have your VUI note how a user should respond. Instead, try something like this:

要明确，但仍要交谈； 让您的VUI记录用户应如何回应。 而是尝试这样的事情：

“Ok, you’re looking for a substitute for milk. What kind of dish is this? Cooked, baked, or fresh?”

“好吧，您正在寻找牛奶的替代品。 这是什么菜？ 煮熟，烘烤还是新鲜？”

3.会话标记 (3. Conversational Markers)

Sprinkle in words and phrases that help orient your users.

撒上有助于定位用户的单词和短语。

Timeline-related markers such as “to get started” or “next” help set users’ expectations regarding duration and subsequent steps. Acknowledging language like “got it” or “sure thing” let the user know they’ve been understood.

与时间线相关的标记(例如“开始”或“下一个”)有助于设置用户对持续时间和后续步骤的期望。 诸如“明白了”或“确定的事情”之类的确认语言使用户知道他们已被理解。

Just like visual cues such as progress bars on GUIs, conversational markers give a user a sense of what they’ve done, where they are, and what’s to happen next. Furthermore, these markers help divide the conversation into more comprehensible chunks.

就像可视化提示(例如GUI上的进度条)一样，对话标记使用户可以感觉到他们做了什么，在哪里以及接下来要发生什么。 此外，这些标记有助于将对话分为更容易理解的部分。

If your VUI responded to your user’s question about baking substitutions for milk by rattling off every answer, it would likely overwhelm the user. Instead, try something like this:

如果您的VUI回答了每个答案，从而回答了用户有关烘焙替代牛奶的问题，则可能会使用户不知所措。 而是尝试这样的事情：

“I found four milk substitutes for baking. Try replacing with yogurt, sour cream, sweetened condensed milk, or finally evaporated milk, mixed half and half with water.”

“ 我找到了四种牛奶替代品进行烘焙。 尝试用酸奶，酸奶油，甜炼乳或 最后蒸发的牛奶代替，并与水混合一半。

Intro your answer with the number of substitutions you found to establish your user’s expectations, then pepper in the adverb, finally, to divide up the information you’re sharing.

向您介绍答案，以找到可以建立用户期望的替代数目，然后在副词中加入胡椒， 最后分割您共享的信息。

4.留下行动项目结束 (4. Leave Action Items for the End)

If you want your user to retain an important instruction, don’t mention it at the beginning of the sentence — save it for the end.

如果您希望用户保留重要的说明，请不要在句子的开头提及它-请将其保存在末尾。

Prioritizing information looks different on visual platforms than on audio platforms. This is because users digest information differently when they hear it and when they read it. While the rule of thumb when designing for GUIs is “put all important concepts first,” the opposite tends to be true when designing for voice.

在视觉平台上，信息优先级在音频平台上看起来有所不同。 这是因为用户在听信息和读信息时对信息的消化方式有所不同。 虽然在设计GUI时的经验法则是“ 将所有重要概念放在首位 ”，但在设计语音时却恰恰相反。

During one episode of his podcast, The Voicebot Podcast, Bret Kinsella talks with Ben Sauer, a conversation designer at Babylon, about user memory as a factor that VUI designers must accommodate. “If you say…a piece of information you want a person to remember at the start of the sentence,” Sauer explains, “then by the end of the sentence, they’ve forgotten it.” Kinsella remarks about journalists transitioning into VUI: “Radio people always knew that…you always put the main point in the end.” For print and television journalists, though, adopting this best practice required a steeper learning curve.

Bret Kinsella在其播客的一集中， Voicebot播客中 ，与巴比伦的对话设计师Ben Sauer进行了交谈，谈到用户记忆是VUI设计师必须考虑的一个因素。 索尔解释说：“如果您说的是……某条信息，那么您希望人们在句子开头记住这些信息，然后，在句子结尾时，他们已经忘记了。” 金塞拉(Kinsella)关于记者过渡到VUI的评论：“广播人一直都知道……您总是把重点放在最后。” 但是，对于印刷和电视新闻工作者来说，采用这种最佳做法需要更陡峭的学习曲线。

Going back to the previous example, one would assume that after hearing a list of four milk substitutes, a user might like for the VUI to repeat themselves. Rather than explaining to a user that they can ask the system to repeat at the beginning of the VUI’s line, where it’s not yet relevant to a user, mention this instruction at the end, like so:

回到上一个示例，将假设在听到四种牛奶替代品的列表之后，用户可能希望VUI重复自己。 除了向用户说明他们可以要求系统在VUI行的开头(与用户无关)之外重复之外，还请在结尾处提及此指令，如下所示：

“I found four milk substitutes for baking. Try replacing with yogurt, sour cream, sweetened condensed milk, or finally evaporated milk, mixed half and half with water. Let me know if you’d like me to repeat these substitutes.”

“ 我找到了四种牛奶替代品进行烘焙。 尝试用酸奶，酸奶油，甜炼乳或 最后蒸发的牛奶代替，并与水混合一半。 让我知道您是否要我重复这些替代品。”

These are my pointers for designers transitioning from screen to voice — what are yours? Give conversation design a shot by designing a skill on Google or Amazon, then share the new tips and tricks you picked up in the comments section.

这些是设计师从屏幕过渡到语音的我的指点-您是什么人？ 通过在Google或Amazon上设计技能，为对话设计做个尝试 ，然后分享您在评论部分获得的新提示和技巧。