What is a voice assistant?

A voice assistant is a technical dialogue system that uses natural language as a medium of communication. In contrast to a text-based chatbot, the conversation with a voice assistant takes place via the spoken word. Voice assistants perform actions and actions on command. The most famous assistants are called Siri, alexa, Google Assistant. It is particularly popular for use on smartphones and smart home devices.

By the way: In case you're wondering how a voice assistant works: A voice assistant is comparable to a chatbot. In a previous article, we took a closer look at how chatbots work: How does a chatbot work?

Where are voice assistants used?

voice assistants are always used sensibly where natural language makes a difference. That means: It does It makes no sense to use a voice assistant in an open-plan officewhen a lot of people want to work there in peace. But if the Environment restricts other ways of interaction, voice assistants show their great strength. A good example of this is driving: The driver usually concentrates on the road in the best possible way. Clicking around wildly on dashboards doesn't serve that purpose. A simple voice command “Hey Mercedes, turn on the light. ” or “Hey Mercedes, I want to listen to Spotify. ” Disturbs significantly less when focusing on the road.

In addition Are there many voice assistants in the home environment again. Google Home as well as alexa are the most well-known representatives. There, assistants make it easier to use the radio, television, lighting and many other household appliances. For example, Alexa can also be used to start garden irrigation or turn on the television. Even more complex queries, such as reading out recipes and answering questions for which you would otherwise have to open Wikipedia, are child's play with these helpers.

In a business environment Do voice assistants help on the phone in customer service. To pre-qualify inquiries and find the right contact person, voice assistants are already welcoming customers from many well-known companies.

What are the challenges of using voice assistants?

The challenges of using voice assistants can be divided into 3 different categories:

• Dialog Design
• Technical challenges
• Ethical challenges

At Dialog Design This is a challenge in terms of content. It is important to determine what goal the voice assistant is pursuing. Should he be able to turn on and off the light or should he rather ask questions such as “Who is Barack Obama? ” answer. From this, it is then derived which use cases must be covered in order to achieve the goal. Each use case is then designed by so-called dialog designers. That's where words make the difference. Professional dialog designers therefore often create a wording guide, which, comparable to a style guide, determines how the language of the digital assistant should be designed. This is how voice assistants differ in branding.

In addition to the pure content challenges Voice assistants are also defined by technology. The following diagram shows the general structure of a voice assistant:

‍

To work, voice assistants need:

1. Einen Speech recognizer (Speech-to-text): A speech recognizer transcribes the spoken word into text. Artificial intelligence plays a major role here.
   ‍
2. A speech generator (text-to-speech system): Under Speech synthesis Do you understand the artificial production of the human speaking voice. A voice generator does just that and receives text and converts it into speech or an audio file. Neural networks and thus artificial intelligence are also used here. As a result, different speaking styles and languages can also be realized.
   ‍
3. A Dialog management system: One Dialog management system (DMS) ensures that dialogs run correctly during a conversation. Voice assistants usually have a so-called internal state, which determines where the dialog is currently located. This allows the system to know how to process the user's response. Dialogue management systems are the subject of active research. In productive use, dialog management systems are often rule-based and therefore not yet trained with the help of artificial neural networks.
   ‍
4. Connections to interfaces: In order for the intelligent voice assistant to be able to process the user's request, interfaces to third-party systems such as CRM or SAP are often required.

There are also various ethical challenges. In particular, the The question of responsibility and protection of data and privacy It is worth mentioning. Who is responsible if a voice assistant does not work as planned? What if voice assistants can make appointments automatically, as in the case of Google Duplex, and the user does not show up for the hairdressing appointment at the end? Who is responsible here? These questions are the subject of current research.