Leading networking multinational Cisco has unveiled new artificial intelligence (AI) strategies for its popular videoconferencing platform, Webex. The announcement made on Wednesday at the company’s annual WebexOne conference, focused on three key areas – audio intelligence, video intelligence, and language intelligence.
The company said it would implement generative AI systems to improve communication and collaboration via video meetings.
Cisco’s AI Can Summarize Meetings and Generate Answers To Queries in Real Time
Cisco launched an all-new Webex AI Assistant that users can communicate with using natural language. Webex users can prompt the AI with questions and it will generate responses in real time. The large language model (LLM) is also able to recommend tone, format, and phrasing changes in Webex Messaging and Sildo – a Q&A and polling platform for meetings and events, and suggest responses based on the context of an ongoing conversation on the video meeting platform.
For those who missed out on meetings, the Webex AI Assistant can summarize entire meetings or missed portions of events by providing easily readable summaries. Meeting summaries are then organized by the AI into chapters and highlights so that users can easily skim through them to find relevant topics. Similarly, the AI can recap and summarize messages and generate summaries for topics discussed in Sildo.
AI Codec Delivers Highest Quality Video and Audio Even When Network is Weak
Another artificial intelligence solution unveiled by the company was the AI Codec, a generative AI computer program that can encode or decode data streams or signals to help Webex users tackle network issues that may plague video conferences.
AI Codec solves network packet loss by allowing for massive transmission redundancy and has speech enhancement functions like background noise removal, de-reverberation, and bandwidth extension, built into Webex that Cisco claims to deliver “unprecedented” audio clarity.
Webex will also apply a machine learning technique called Super Resolution to improve video quality. The generative AI tool will deliver high-definition audio and video during meetings, regardless of the bandwidth conditions.
According to Cisco engineer Keith Griffin, Webex’s AI Codec uses up to 16 times less bandwidth than any industry-standard audio codec. The system is fully optimized for speech and requires just a fraction of the bandwidth of other codecs, said Griffin. This means that the AI gives out the highest possible quality of speech even when bandwidth is extremely limited.
RMMs can Sense if a Person has Abruptly Left the Meeting and Take Down Notes
Cisco also introduced a generative AI tool called Real-Time Media Models (RMMs) that is said to further enhance audio and video quality in Webex. RMMs are able to produce multiple audio and video outputs for multiple media streams, including people and object recognition and action analytics like movement and gestures.
RMMs in Webex enable audio and video channels to be used as context signals in text-based operations such as meeting summaries and highlights. The unique tool can fuse together AI for text, audio, and video, giving Webex users the unique ability to gain real-time insights into events and meetings. For instance, Webex now recognizes if a participant has moved away from the meeting and it will automatically take down notes to help that person get up to speed with the points they missed out on.
Commenting on the company’s new AI capabilities, Griffin said while the whole world is discussing LLMs and generative AI in text, these tools have abilities that are undiscovered. While text is a key component of Webex, the platform is more focused on other inputs that can be generated by machine learning models like speech, video, and computer vision.
Cisco highlighted that while most LLMs are better suited than others for specific use cases and training, the Webex AI Assistant will be using a combination of “best-of-breed models”, which include commercial, open source, Cisco-proprietory, and select customer-based models, to deliver the best possible meeting experience to customers.
Webex’s new generative AI capabilities are in various stages of availability and development. The company plans to ship the new version of the videoconference software before the end of 2023.