For those who want more depth,  here is an Audience thought item on voice....

Audience eS700 Series – The Analyst Point of View

Written by: Linley Gwennap

Audience regularly works with leading industry analysts, like Linley Gwennap of The Linley Group, to keep them informed on our latest news and to gain unique third party perspectives on the industry that can help shape strategy and business decisions. When Audience introduced our eS700 series, we worked closely with Linley Gwennap on his whitepaper: “Always Listening, Always On: Advances in Sensory Processing.”  Below is an abstracted version. To read the entire whitepaper, go to Linley’s website.

Always Listening, Always On: Advances in Sensory Processing

Smartphones use sensors and voice input to become aware of their surroundings and to interact more naturally with their users. Smartphone makers continue to develop more advanced capabilities, adding new types of sensors and more sophisticated voice functions. If implemented improperly, these changes can greatly reduce battery life.

The first smartphones tried to bring the PC experience into the palm of your hand. Since then, smartphone makers have raced to add more capabilities, helping their devices to be aware of and react to their surroundings. These capabilities depend on an expanding array of sensors that react to gravity, motion, and light.

Voice has become a central function of high-end smartphones. Virtual assistants respond to natural-language commands, performing information searches and controlling aspects of the device. Developers are now taking these functions to the next level. By comparing sensor data to existing motion signatures, software can recognize gestures, count the number of steps the user has taken, or analyze the effectiveness of their exercise.

Smartphones Adopt Sensor Hubs

High-end smartphones continue to add sensors, including a magnetometer (compass), barometer (pressure), ALS (ambient-light sensor), UV (ultraviolet-light) sensor, and infrared proximity sensor. As their costs fall, these sensors will migrate into mainstream phones, just like the gyro did.

Managing all of this sensor data is becoming complex, particularly for applications that continually measure motion, such as a fitness app. Using the main processor to monitor the sensors would be power intensive. High-end phones use a “sensor hub” to coordinate this data, waking the main processor only when necessary.

Voice Processing Becomes “Always On”

Voice is another form of sensory input. Originally voice command functions required user-specific training, but modern processors can implement speaker-independent voice recognition. Sophisticated services, such as voice search and virtual assistants, need large vocabularies; these services often implement voice recognition in the cloud, sending the voice sample to a distant data center.

A device that is “always on, always listening” solves this problem. This technology allows the user to create a verbal phrase that will wake the device. The phone can then accept voice commands such as “set alarm for 10AM,” all without ever touching the device. Other sensory-based gestures, such as wrist flicks, can reduce the time needed to open an application.

Always Listening, Always Power Efficient

Most phones today implement voice commands on the main, power hungry CPU. As with other sensors, the key to “always listening” is to offload this function from the main CPU on to a smaller, power-efficient processor that creates minimal impact on battery life. The new eS700 chip from Audience uses just 0.6mA current draw from the battery when “listening,” meaning that turning on this feature will reduce the battery life of an average phone by only 15 seconds.

The eS700 series combines a processor and DSP capability on a single chip. A DSP uses less power than a standard CPU for voice processing, because it has been optimized for that function. Like a microcontroller, the Audience chip also integrates memory, allowing it to store the voice buffer and related data. Keeping this data on chip reduces both power and cost, compared with external memory.

Emerging Applications

Although smartphones are the initial focus for always-on features, this technology applies to other types of devices. Always-listening technology is applicable to tablets, laptop PCs, smart TVs and other consumer-video products, automotive in-dash systems, and wearables. As these markets develop, consumers will discover a new, more natural way of interacting with their devices.

Conclusion        

Enabling this sensory processing functionality requires equally advanced hardware. No matter how powerful the application processor, it cannot perform these functions without quickly depleting the battery. For maximum power efficiency, sensor and voice processing must be offloaded to a separate processor. This approach, combined with intelligent power-management software, will maintain battery life even when the phone is “always on.”

With integrated memory and DSP capability, the Audience eS700 series are low-power processors that are well suited for always-listening, always-on devices.

For the casual reader.  For the MAJOR application solution, Indoor location leaving the MIC on for YA won't do ananything as it won't know where you are.  But it is an alternative input to a wearable to gesture and so its nice that we will have it in the mix and match algo store......