Artificially intelligent ad-break prediction

A method for use by an artificial intelligence (AI) system including a computing platform having a hardware processor and a system memory storing a software code, the method comprising:

• receiving, by the software code executed by the hardware processor, a media content including a video component and an audio component;
• determining, by the software code executed by the hardware processor, whether the media content is an ad-slugged media content that includes pre-determined ad-slugs or is a seamless media content that does not include any pre-determined ad-slugs;
• detecting, by the software code executed by the hardware processor, a plurality of black video frames of the media content;
• if the media content is determined to be the ad-slugged media content that includes the pre-determined ad-slugs, applying a first method of ranking ad-insertion points, the first method comprising:
• detecting, by the software code executed by the hardware processor, a plurality of silent video frames of the media content;
• and
• predicting, by the software code executed by the hardware processor and using a first plurality trained machine learning (ML) models and based on the plurality of black video frames and the plurality of silent video frames, one or more candidate ad-insertion points for the media content and a respective probability score associated with each of the one or more candidate ad-insertion points to provide one or more ad-break predictions for the media content;
• and;
• if the media content is determined to be the seamless media content that does not include any pre-determined ad-slugs, applying a second method different from the first method, the second method comprising:
• performing, by the software code executed by the hardware processor and using an ML model trained using contrastive learning,, a first plurality of evaluations of fade-to-black transitions among sequential video frames of the seamless media content;
• performing, by the software code executed by the hardware processor based on comparison of decomposed audio signals each sampled from a respective video frame of the seamless media content, one or more second evaluations of audio continuity across respective one or more sequences of the plurality of black video frames;
• predicting, by the software code executed by the hardware processor and using a second plurality of trained ML model different from the first plurality of ML models and based on the plurality of black video frames, the first plurality of evaluations, and the one or more second evaluations, one or more candidate ad-insertion points for the media content and a respective probability score associated with each of the one or more candidate ad-insertion points to provide one or more ad-break predictions for the media content;
• and
• outputting, by the software code executed by the hardware processor, the one or more ad-break predictions for the media content.