(Credit: Ole_CNX / Getty Images)
Cisco is turning to AI to solve some of the most common issues with video calls, such as blurry video and choppy audio.
The company announced a suite of new AI products for its Webex video-conferencing software today that it says will "achieve new levels of audio and video quality."
AI Audio Codec will fill in gaps in speech when there is a spotty Wi-Fi connection or a disruption on a call—when leaving a car, for example. It does this by using "far less bandwidth to transmit speech vs. the industry benchmark," Cisco tells PCMag. This allows Webex to create "redundant" copies of the audio feed and serve up a new one if the primary gets lost.
"In the future, we will also be able to use generative AI to reconstruct the audio in case the dropped audio frames are unavailable," Cisco says.
While call participants may appreciate more consistent audio, this type of real-time deepfake creation could raise ethical issues. Cisco says its approach to AI aims to realize its "significant promise while adhering to standards for transparency, fairness, accountability, privacy, security, and reliability."
Webex's new video technology, dubbed Super Resolution, works similarly to the audio. "We take a feed and downscale it to save compute resources as the packet travels through the network," Cisco says. "Then, when it reaches the user, it is transformed back to 1080p—the end result is a higher resolution than the original video."
Following competitors like Zoom and Google Meet, Cisco also unveiled an AI Assistant that can summarize meeting notes. Ask it to "catch me up on the 15 minutes I missed in the meeting," or for suggestions on how to improve the tone of a message to a colleague or customer. These features are "in various stages of availability" and will start rolling out before the end of the year.
Cisco plans to combine multiple large language models to enable these capabilities, including "commercial, open source, Cisco-proprietary and select customer models."
In the future, its AI products could help distill insights from human behaviors. For example, an AI-generated alert might be able to "inform meeting participants if someone left the room when a critical topic was discussed or if a dramatic pause in speech is made indicating contemplation."
The use of AI in video calls has caused some controversy. Earlier this year, Zoom found itself in hot water after updating its terms of service to say its AI products could train on user data. After backlash, it rolled back the policy, but has continued to experiment with an AI companion to summarize meeting notes.