Voice Control — What‘s Coming?
Technology advances rapidly. In the coming years, voice control will be expected in homes, offices, cars and everywhere in between. To create the ultimate smart home experience (especially in the custom channel), we saw an evident need to manufacture our own hardware that incorporates a microphone array, a whole home processor, sensors, a capacitive touch dial to augment voice commands, and much more. With our microphone also acting as the processor, Josh Micro enables room awarenesses and a network of communication between each device. If one has an issue, goes down, or is slow, there is a redundancy failover to ensure the system will stay operational.
Josh’s artificial intelligence platform is designed to not only revolutionize how voice control is working in the home, but also the installation model, especially the historic need for a dedicated processor in a rack. Singularly focused on home control, Josh outperforms other voice platforms with its intuitive interface and flexible natural language processing abilities. With dynamic voice control with dedicated hardware built for the machine learning to come, the Josh platform is a big step towards homes learning routines as well as pro-actively taking action based on historical routines and the time of day.
So what sort of developments do we expect to see in voice control in the future?
One of the most requested features for voice control is voice recognition to easily distinguish between everyone’s individual voices. This will create a more personalized experience for each unique user to say commands like, “listen to my morning playlist,” or by simply saying “party time” to set the environment to their party preferences. This will also act as a security feature to prevent the microphone from responding to house guests, workers and even children.
Josh Micro processes everything it can locally on the device, located in the home. Because of the required processing power, the speech-to-text (ASR) portion of a command has to be sent to the cloud to convert the audio to a text string. The text string is sent back to the local device to parse out “what you actually want to happen” based on what you said. This is more secure compared to processing voice commands in the cloud.
In the near future, technology will improve so that we can also quickly and reliably process the speech-to-text locally, to offer an even more secure system that doesn’t require an external network connection.
Voice control and specifically Automatic Speech Recognition (ASR) have become increasingly accurate over the years, with a 95% success rate for the English language. However, English is the only language that can fluidly be used for voice control. This leaves out 80% of the world’s population.
Heavy accents are also challenging to accurately translate for the ASR, with frequent mis-hearings and incorrect transcriptions to a text string. ASR will continue to improve so that people speaking in any language or with an accent can naturally control their homes via voice.
In order to build a completely automated smart home, sensors should act as conditional triggers for scenes. This will enable homes to better decide what to do on their own. For example, if a motion sensor activates when you enter the Family Room after 6pm during the week, your home could know based on previous patterns that you want the lights dimmed to 30%, the shades closed and the TV turned on to where you left off on Netflix. Sensors will replace the traditional need to create custom scenes for various events throughout the home.
True Artificial Intelligence
We believe voice control is only the first step to a more human-like system. Instead of asking to turn on the lights, your house should know that the lights should be super bright at noon and dim at 11pm based on historical data and patterns. If you come home every weekday between 6 and 8pm, and you open the garage door, close it, and then turn on the lights in the Kitchen, your home should learn. If you come home during that time and open the garage, the remaining sequence of events should automatically happen without asking.
Down the road, users will have an opt-in automation option for their home to understand their daily routines, enabling the house to proactively set the scene without user input.
It is important to know what integrates best together to create a natural user experience. Some products are better optimized for voice control, with open APIs, IP control, etc. Our goal is to be completely brand agnostic, acting as the AI layer to cohesively bring all smart products on the network together for fluid voice control. Regardless of the manufacturer, Josh wants to quickly auto-discover what’s on the network and control it all — by doing a lot more than just voice.
The types of hardware available for you to talk to your home are near-field (smartphones) or far-field microphones (Josh Micro, Amazon Echo). Both types are visible devices with a known presence in the home. In the coming years, the technology will evolve to be invisible, discrete microphones, that will seamlessly blend into any environment. Interior designer approved, these could be hidden in-ceiling or incorporated into your kitchen counter’s granite with a Trufig-like concept.
Voice control everywhere
Voice will naturally spread beyond the home into all of our environments, and this extension of the technology will create a fluid user experience. For example, while driving, users will be able to tell their car to preparing the house for their arrival. As they pull into their garage, it will be as if they never left.