ABI Research is forecasting major growth in the audio chipset market. The firm’s latest report predicts that tech manufacturers will ship more than 2 billion devices with a chipset designed solely for ambient sound or natural language processing in the leadup to 2026.
According to ABI Research, much of that growth can be attributed to recent advancements in audio chipset design. As it stands, many of the top voice assistants (including Siri, Google Assistant, and Amazon Alexa) operate in the cloud, and cannot run without an internet connection. That limits the performance of those systems in certain situations, and makes voice technology less appealing in industries that have privacy or security concerns.
Dedicated audio chipsets, on the other hand, enable voice and sound processing at the edge. That means that the systems will still work while offline, and that they can be integrated into the design of the device itself to provide a higher level of security.
That transition to the edge audio environment is already underway in the consumer sector. Apple revealed that Siri would have some offline utility earlier this year, while Google is reportedly developing a new Tensor System-on-Chip (SoC) to enhance its face and speech recognition capabilities. The first Tensor Processing Unit debuted all the way back in 2016.
In the meantime, ABI expects ambient sound processing technology to be popular in industrial and manufacturing facilities. Sensors can analyze the sounds of machines to watch for signs of early wear and tear, giving operators the chance to carry out predictive maintenance and fix their equipment before a machine breaks down entirely.
“NLP and ambient sound processing will follow the same cloud-to-edge evolutionary path as machine vision,” said ABI Machine Learning and AI Principal Analyst Lian Jye Su. “At the moment, most of the implementations focus on simple tasks, such as wake word detection, scene recognition, and voice biometrics. However, moving forward, AI-enabled devices will feature more complex audio and voice processing applications.”
ABI noted that many chip manufacturers are forming partnerships to improve their technology. For example, Syntiant and Renesas are working together to build a processor that supports both computer vision and voice applications, while Qualcomm has allied with the language processing start-ups Audio Analytics and Hugging Face. The tech giant has also released chipsets designed for use in smart audio devices.