InfoComm 2022

The Triumphant Return to Vegas

This is the first InfoComm CX has been able to attend since 2018, and it certainly did not disappoint. Now occupying the shiny and huge new West Hall as well as the old North Hall at the Las Vegas Convention Centre, the 2022 edition has come back with a larger footprint, more exhibitors, and more attendees. It was a joyous reunion of the global AV industry, even if we APAC folk were a bit thin on the ground.

The stands were busy, and the audio demos rooms were packed. If you were worried about COVID, it was the wrong place to be. On the last day of the show, InfoComm sent out a warning from Clark County (where Vegas resides) that rates of transmission had dramatically escalated and everyone should wear masks inside. I was already getting on a plane back home. Having spent three happy days cheek-by-jowl with the international AV community, I was past caring.

The big product and tech news from the show came from two unlikely sources; a video conferencing bar from Biamp and a captioning system from Williams AV. What did they have in common? They’d both been trained by AI.

Biamp’s Parlé video conferencing bars have beamtracking microphones, great speakers, and a nice camera, but the absolutely astonishing thing about them is their AI-trained noise reduction. I sat in on the demo, and there is no audio tech on the planet that can do what they did. An AI was taught to distinguish between human speech and ‘not human speech’ and applies an aggressive algorithm to the audio in real time. The live demo saw the remote participants crunching chip packets, using drumsticks on the table, and even operating a drill, and the AI just took it all out, while we could still hear them talking.

Williams AV real-time video captioning system Caption Assist can transcribe live speech into 73 languages with up to 94% accuracy. Williams teamed with Google’s Neural Network AI to train it.

The implications for AI in AV are huge. If Biamp can use an AI to train DSP to remove speech, we can also use it to train DSP to emulate classic audio gear like pres and mics, or achieve the “perfect” vocal tone. It could also tune entire speakers systems. If it can be trained to turn audio into captions, it can be trained to alter video in almost any way we can imagine.

We captured dozens of product videos which we’re still editing together, just three of which are in today’s CX News email (subscribe here if you’re not already receiving it) and we’ll be posting them all to social and subsequent newsletters as they’re done. There will be a full InfoComm 2022 feature in CX Magazine July, which you should subscribe to in print – they will be way more in the mag than we’ll ever put on the web!