Voice bots are very similar to chat bots, which have been around for a long time and support customer service as a text-based dialog system on websites or in messenger services. Requests or questions are typed in using the keyboard and answered by the bot in the chat. In the same way, a bot also works by voice, i.e. by voice in a voice dialog. The ChatBot principle is thus transferred from the written to the verbal level. This enables completely new dimensions in a multimodal dialog with companies: Communication in natural language via telephone, smartphone, tablet, PC or even smart speakers and devices such as Alexa, Google Assistant, or Siri with their apps.
The voicebot uses AI-based Natural Language Understanding technology (NLU) sometimes called SLU = Spoken Language Understanding technology, to recognize the spoken words. This works with very high recognition rate and interpretation accuracy even for complex requests and multiple concerns in one sentence. Alternatively, what is said can first be transcribed via Speech-to-Text (STT) and analyzed downstream.
The identified request is compared with existing cases in knowledge databases, with the help of sophisticated AI mechanisms. This way the caller's request (also called intent) is interpreted and a most suitable and helpful answer selected. To achieve maximum quality, the voicebot has to be trained domain-specifically in advance. In addition, company databases are queried to provide information on customer-related information such as account balances, credit lines, credit balances, etc. The new generation of AI-supported voicebots is also able to optimize answers independently using trial and error.
The voicebot plays out the answer via text-to-speech (TTS/speech synthesis), the speech quality of which is now barely distinguishable from sound recordings of a real human voice.
Of course, all this works with minimal, barely noticeable delays to achieve an experience like talking to a human.
Ideally, a dialog between human and machine is created in natural language, where the voicebot reliably recognizes the caller's request, automatically answers their query and, if possible, concludes the case or connects them to the right person or department in customer service.
Speech is the most immediate and simplest form of human interaction and offers a high degree of accessibility. In addition, a mobile or landline phone is available almost everywhere. That's why voicebots are particularly well suited to noticeably improve a company's customer service.
Voice Portal, Voice Computer, Voice Dialog System, Conversational IVR