Voicebot is an AI-based interactive machine with which human-machine conversations are possible today. These voicebots are trained to understand human speech and can have near human-level interactions.
But voicebot over the phone is a different concept altogether. With voicebot over the phone, businesses can make phone calls to customers which means a machine will call a customer and engage in a conversation, all these in a click of a button.
It is even more interesting to know that a voicebot is capable of calling 1000s of people within 5 minutes.
Bluntly put this is machine calling human and interacting with them
Can we say a voicebot is 900% more efficient than a customer agent?
In a way, this is an upgrade to voicebot which are already available in the market (Siri, Google Assistant and Amazon - Alexa all these are voicebots).
The upgrade to voicebot over the phone will definitely open up new unexplored markets where machine intervention was limited or was not possible previously.
Few areas where a voicebot over the phone is already deployed:
Customer qualification in sales
Customer engagements like feedback collection
These are just a few examples, a day will definitely come when a fully functional robot will interact with you on the topic of your preference.
Qualyon is a Voicebot-Over-Phone
Qualyon is not available on a webpage or a smartphone app (at least not right now). This is just the specialization we have. We believe there is a large market for reaching people over the phone.
But this is also a very difficult problem to solve. This brief article talks about our experience in solving this problem.
What's the problem with Phone audio?
When a call is dialled, a person speaks on a phone and phone converts your voice into an electrical signal, which is then transmitted via radio waves to the nearest cell tower you know the whole story.
Phone systems work on 8000 samples per second which results in low-fidelity voice. This makes it quite difficult for an automatic speech recognition system (ASR) to understand.
A Voicebot on a webpage has the advantage of a high voice sampling rate, typically 44100 samples per second. But that's not all. An Alexa or Google Echo device has multiple mics. This is used for noise cancellation so that a better voice capture can be done. Adding to all these are network interferences because of which audio signal can be lost permanently.
How does Qualyon handle this?
One can't break the laws of physics. Accurate handling in this situation requires very high-level AI processing. Due to low voice fidelity and extraneous noise, most of the times the voice is not correctly transcribed. Understanding in a sub-optimal acoustic environment requires application of AI techniques and we have employed two kinds of techniques at Qualyon:
Anticipate the conversation: Create the correct set of expectations for what to listen for. Just as it happens for us, if we know we are closing our conversation, it is more likely the caller is trying to say "bye", instead of "buy"
Redirect the conversation with smart responses: Control the conversation by asking smart questions that cause the caller to answer in a specific way. Open-ended questions can result in users saying whatever comes to mind, and that makes it difficult to predict and say an appropriate answer.
Qualyon uses advanced techniques to detect such errors. Errors in a live system are monitored for improvement. It is measured as Voice Recognition Efficiency. This will act as a measure to detect how efficiently the voicebot is able to handle and answer the question with relevant responses.
In 2020, having a voicebot in businesses will playout as a big business differentiator. We are entering an era where conversational voicebots will come into the mainstream and enhance the customer experience.
Companies who have or will soon establish a voice presence will start to reap the business benefits over laggards, much like what happened with the web and mobile.