We noticed you are using Internet Explorer.

Voice-controlled devices like TV remotes have taken the world by storm. Learn more about the technology that powers them.

From smartwatches and smart home hubs equipped with digital voice assistants to voice-activated car ignitions, the number of consumer products with voice processing capabilities continues to climb. According to NPR’s Winter 2019 Smart Audio Report, 54 percent of adults in the United States have used voice commands, and of those who use voice, 24 percent use voice commands daily. That means nearly 13 percent of all American adults are using some form of voice command every single day.

Voice processing is usually accomplished using one of two approaches: cloud or local (commonly referred to as “voice-at-the-edge”). Exemplified by devices like Amazon’s Alexa, cloud-based voice processing involves an IoT device picking up an audio command from a user and delivering it to the cloud, where it is processed and then sent back to the device — typically in the form of an action for the device to take. Conversely, local voice processing brings the entire process closer to where audio data is being produced and gathered, utilizing edge devices’ compute power instead of the cloud’s.

While these approaches are most commonly associated with devices like smart speakers, the same underlying mechanics apply to voice-enabled remote controls. Local voice processing in particular delivers numerous benefits to consumers and remote control manufacturers alike, but there are still opportunities to improve on this technology. Here is everything you need to know about voice-at-the-edge processing, plus the key ways Telink is driving innovation in this space:

Voice-at-the-edge Explained

Typically, the voice recognition software that enables edge processing uses layers of sophisticated algorithms to detect, parse, and mimic human speech, and remote servers execute the bulk of this processing offsite. A user speaks a trigger word, their voice assistant picks it up, audio data streams into the cloud, the data is processed and recognized in the cloud, and a response is formulated that is then sent back to the user’s IoT device. However, if hundreds of thousands of devices are transmitting data at the same time, processing speed will suffer, and companies will have to compensate by spending millions of dollars on additional cloud instances.

Voice-at-the-edge processing solves this dilemma by serving as a local source of processing and storage for many systems. Voice-at-the-edge only sends relevant data through to the cloud instead of all data, which reduces bandwidth needs and latency. In addition to using cloud bandwidth more efficiently and improving voice-enabled remote controls’ response times, processing voice data locally often results in:

• Less expense for manufacturers: Local processing keeps bandwidth, electricity, and hardware costs to a minimum by reducing transmissions of data from devices to the cloud. Further, cloud-based voice recognition APIs can cost up to $4.00 per 1,000 API calls, but they are not necessary with voice-at-the-edge.
• Better energy efficiency: Instead of constantly collecting all words that are spoken in their vicinity, remote controls utilizing local processing use a small amount of resources for “wake word” detection while the rest of their systems remain idle. This saves a great deal of energy, ensuring consumers’ batteries last longer.
• Greater privacy: With voice-at-the-edge processing, a user’s voice data — which could be used as an identifier — never leaves their local network. Security, privacy, and compliance concerns are nearly eliminated.

Bluetooth^® Low Energy: A Proven Standard for Voice-Enabled Remotes

Over the last several years, a growing number of set-top box and OTT box manufacturers have been equipping their Android-based solutions with voice capabilities through the inclusion of features like the Google Voice application. To enable users to utilize these voice capabilities, many manufacturers are replacing infrared remotes with remotes built around the Bluetooth^® Low Energy communications protocol.

Bluetooth LE not only delivers improvements on standard remote functionalities — better energy efficiency, non-line-of-sight operation, and so forth — but offers a proven method for local transmission of voice data.

Telink’s TLSR827x Bluetooth LE and multiprotocol SoCs are designed to support superior audio and voice performance, and have achieved a 100 percent pass rate in tests against Google’s and Nuance’s audio specifications. Our SoCs are compatible with Google Voice for Android TV (Android 8 and 9) and our voice-enabled remote control software development kits help manufacturers deliver the voice capabilities they need to satisfy the latest, most groundbreaking consumer trends.

Recent Innovations Driving Voice-at-the-Edge

The development of communications protocols like Zigbee RF4CE and Bluetooth Low Energy (LE) Audio promises to take local transmission of voice data to new heights.

Zigbee RF4CE is a low-power solution that enables low-latency two-way communications for a wide range of voice-enabled products, from keyless entry systems to remote controls. A full-featured Zigbee RF4CE remote control will often have a digital PDM microphone that records voice data through an audio inlet at the top of the remote. Since PDM is in digital format, users can hold the microphone in the remote a reasonable distance away from the hardware codec without negatively impacting audio performance. Other notable features include longer battery life, longer range of use, robust “Find My Remote” features, and non-line-of-sight operation.

Bluetooth LE Audio is the next generation of Bluetooth audio that enhances performance and supports all of Classic Audio’s use cases, but adds newer features, as well. This technology was originally developed to help people who are hard of hearing, but Bluetooth LE Audio’s enhanced listening capabilities make it an excellent addition to voice-enabled remote controls. Bluetooth LE Audio uses a low-complexity communication codec (LC3) to offer higher audio quality at lower power and data rates. In fact, extensive testing has proven that LC3 provides significant improvements in audio quality over the SBC codec that is included with Classic Audio while operating at a 50 percent lower bit rate.

What is more, the LC3 codec allows designers to integrate a complete hardware and software solution for voice processing into their voice-enabled devices with minimal energy consumption, which translates into minimal cost. With LC3, remote control manufacturers gain greater flexibility and do not have to choose between design, key product attributes, audio quality, and power consumption. Innovations like these help manufacturers deliver voice-at-the-edge processing at scale, but there are always opportunities to improve and further reduce costs — and Telink can help.

The Telink Difference

Telink is a leader in chip innovation, including low-cost, energy-efficient, voice-enabled chips for remote controls. Local voice processing typically shifts some costs to either consumers or remote manufacturers, but intelligent chips like Telink’s offer better voice transmission and processing at a lower cost, eliminating any sort of trade-off.

Every remote is different, but Telink’s product line can meet the requirements of every single one. We power millions of remotes across every device class — no matter your I/O, power requirements, or feature demands. We integrate everything you need into a small, cost-efficient package and support all popular protocols — even the ones you might need in the future.

Our chips power remotes for some of the largest smart TV, set-top box, and OTT box manufacturers in the world. We have developed multimode solutions that deliver exceptional performance and unparalleled flexibility at an unbeatable price point. To learn more, contact us today.