Artificial Intelligence on Developing Diffent Algorithms
April 6, 2023
In recent days, the transnational technology company from the United States Apple has acquired the WaveOne organization, which is dedicated to developing Artificial Intelligence algorithms to automatically compress videos.
Although Apple hasn’t officially confirmed the purchase, WaveOne’s former head of sales and business development, Bob Stankosh, made it known in a LinkedIn post.
“After almost 2 years at WaveOne, last week we finalized the marketing of the company to Apple. We started our journey realizing that machine learning could change the planet. Apple saw this potential and jumped at the chance to add it to its technology portfolio”
Stankosh described in the post. With recent standard video compression and decompression algorithms, compression is performed by the content provider platform, for example, YouTube servers. While the decompression is the responsibility of the end customer.
WaveOne’s main innovation was to generate a “content-aware” video compression and decompression algorithm. Taking advantage of scene detection by artificial intelligence, the technology is able to “understand” a video frame. In this way it allows, for example, faces to be prioritized over other resources in a scene to save bandwidth when playing back content.
Before starting the company, Bourdev was a founding member of Meta’s AI research division. In addition, both he and Rippel worked on the Meta accessories responsible for content moderation, visual inquiry, and categorization of ‘feeds’ on Facebook.
The compression technology developed by WaveOne is also resistant to sudden interruptions in connectivity. In other words, it can make a “best estimate” based on the bits it has available, due to which, once the bandwidth is suddenly restricted, the video does not freeze, but shows less detail.
The new algorithm created by WaveOne is capable of minimizing the size of video files by up to half, obtaining superior results in quality and fluidity in the most complicated scenes. With this purchase, Apple seeks to create a more efficient streaming platform. Small improvements in video compression have the potential to save on bandwidth costs or allow services like Apple TV+ to deliver higher frame rates and solutions.
WaveOne was founded in 2016 to generate video codecs based on Artificial Intelligence. Its “content-aware” video clip compression technology could interpret a video clip frame and prioritize certain resources over others. Consequently, it would “understand” that faces and texts are elements to prioritize. This would even save bandwidth.
It is also worth remembering the acquisition of Apple that has also been significant for the company, which would be acquired by Voysis for Siri. Siri has been one of the first voice assistants to reach smartphones. Acquired by Apple in 2010, it made its first appearance a year later, in October 2011, with the launch of the iPhone 4S. Since then, the company has been integrating it into other of its devices such as the iPad, Apple Watch, computers with macOS, Apple TV, or the HomePod smart speaker.
However, in recent years other competing voice assistants have outperformed Apple’s initiative, mostly in languages other than English. Its main rivals are Google Assistant and Amazon Alexa, both integrated into smart speakers, smartphones, televisions, and tablets, which in languages like Spanish tend to solve user requests. Even though in 2018 they introduced Siri Shortcuts to increase the assistant’s abilities, Apple has to go a step further to catch up with its participants.
For a product with these properties to be a success, it should be able to understand human language in the best possible way, and it does not respond with stability only to pre-established voice commands in which if you use a synonym it stops working. Voysis launched a platform for digital voice assistants with which to better understand the natural language of individuals.
What is Voysis?
Based in Dublin, Ireland, Voysis had focused on improving the assistants included in online shopping apps so that the program would be able to more accurately respond to users’ voice commands. For example, the technology was capable of minimizing the query results with phrases such as “I need a new LED TV” and “my budget is $1,000.”
The system provided by Voysis is based on Wavenets, a procedure that uses AI to produce a more human-computer discourse. It was first created by Google’s DeepMind in 2016. Voysis co-founder Peter Cahill mentioned in 2018 that his company was able to compress its system so much that, once doing a job, it was able to use only 25 megabytes memory. This greatly enhances its use on smartphones without internet access.
Although Apple has not revealed which planes it has after the purchase of Voysis, due to its properties it could be feasible for the company to use the technology to improve Siri. A few months ago, it was learned that Apple was also achieved with Xnor.ai, another startup specializing in AI (artificial intelligence) whose technology had been used in Wyze home stability cameras to detect individuals.
Another of the current purchases of those from Cupertino is Dark Sky, an application for smartphones with iOS that provides planned information. The truth is that the native iPhone Weather app has hardly introduced any changes for a long time, so Apple’s goal may be to improve it.
It is expected that Dark Sky accessories will become part of Apple in its entirety here, we assume, it will work on improvements for the Weather app that we could find installed on all our iPhones. The improvements that can be expected from this purchase (perhaps in the next WWDC) range from a redesign of the app to more descriptive and strict information, without a doubt, good news.
The purpose of the team was constantly to provide the best information while respecting privacy. In the same way that they expose the accessories “there is no better place to do it than in Apple” and they remain quite happy with the possibility of reaching many more people than they could reach as an independent app.