With the advances in artificial intelligence (AI), there have been a lot of buzzwords going around: bot, chatbot, virtual assistant, voice assistant, and voicebot along with technologies used to build them, such as Speech-to-Text, Natural Language Understanding (NLU), Speech-to-Intent, and Text-to-Speech. The lack of standardization of terms makes things difficult even for AI experts. Let’s unwrap what they mean.
What’s a chatbot?
A chatbot is an automated software powered by AI to simulate human conversation by matching keywords. A chatbot takes the keywords from users’ messages, mostly questions, figures out users’ intentions via a natural language understanding (NLU) engine and finds out the best possible answer. Chatbots are pre-programmed with tens or hundreds of intents, variations of expressions (i.e. how users may interact with it) and pre-recorded replies.
What’s a voicebot?
A voice chatbot, or voicebot in short, works quite similarly to chatbots. They also simulate human conversation by matching keywords. A voicebot takes keywords from users’ utterances (i.e. spoken words or statements) instead of text messages. To understand users’ intents from utterances, voicebots use Automatic Speech Recognition (ASR) technology. In most cases, a voicebot first converts voice to text via Speech-to-Text (STT) and then runs NLU to understand the user intent. For context-aware voicebots, such as IT Help Desk IVRs, due to the high accuracy, Speech-to-Intent is preferred instead of running STT and NLU engines separately.
You can build a voice assistant using Porcupine Wake Word, Orca Text-to-Speech, Rhino Speech-to-Intent, Leopard Speech-to-Text, and a large language model of your choice. Further, you can personalize the experience with Eagle Speaker Recognition!
What’s a virtual assistant?
A virtual assistant is an AI-powered software that assists humans in performing tasks. A virtual assistant also takes keywords from users’ messages, figures out users’ intentions via an NLU engine and performs a task. A chatbot is a version of a virtual assistant that answer questions of a user or help them navigate menus.
What’s a voice assistant?
A voice assistant is a version of a virtual assistant that takes user inputs in utterances, i.e. voice commands. A voicebot is a version of a voice assistant. With the increasing popularity of voice-based virtual assistants over text-based assistants, virtual assistants and voice assistants are now used interchangeably.
Where does the confusion come from?
It’s not wrong to say that chatbots and voicebots are different versions of the same core technology, conversational AI. Conversational AI is the discipline that allows humans and machines to engage in a conversation, written or verbal. In the past, both chatbots and voicebots were “dumb”. Users were asked to choose among given options and the accuracy of voicebots was poor. Now with the advances in voice AI, voicebots, voice assistants and virtual assistants are mostly used interchangeably.
When to choose a chatbot or a voicebot?
It depends on the context. As conversation design expert Erika Hall mentioned in her recent interview: A user of a banking application would choose a chatbot if they’re on public transportation, and a voicebot if they’re in the middle of cooking for a completely hands-free experience.
A chatbot wouldn’t be preferred for tasks such as making comparisons between products, in e-commerce in general, checking stock market updates, playing music, or managing smart devices as a user would have access to Graphic User Interface (GUI) to make that changes much faster than typing. Hence, chatbots are mostly used for triaging customer support queries. Enterprises have started complementing chatbots with voicebots as talking is 3x faster than typing. Plus typing is not preferred among certain age groups, people with disabilities, or small screens.
What to use: voicebot, virtual assistant or voice assistant?
The most known assistants in the market such as Alexa or Siri are known by their well thought and experimented names, not as a voicebot, or a voice assistant, just Alexa... Even changing “Alexa” to “Hey Amazon” affects user perception. To avoid confusion, invest in your brand with a custom branded wake word. Let users call your product by your brand name, rather than a bot or assistant! Picovoice’s Free Plan allows users to train custom branded wake words in seconds and deploy them to the platform of choice.
What’s next?
Train your models and select your favorite SDK to start building your first voice feature in minutes - for free!
