Over the last decade I've been asked many times whether I think blogging will die, or will it be replaced by something else. Every time I was asked this question I replied that I believe the only way it could change is if how we consume content changes. As long as we humans like to read, listen and watch, and we use our eyes and ears to do so, I couldn't imagine anything changing. Unless technology reaches the point where we can download content directly to our brains like in the Matrix movies… which I don't see happening anytime soon! While my basic thesis remains the same – we continue to use our eyes and ears to read and listen to content – for the first time in a long time there is something on the horizon that is going to cause a fundamental shift.

In recent years, deep learning has surpassed traditional approaches to the problem of singing voice separation. The Wave-U-Net is a recent deep network architecture that operates directly on the time domain. The standard Wave-U-Net is trained with data augmentation and early stopping to prevent overfitting. Minimum hyperspherical energy (MHE) regularization has recently proven to increase generalization in image classification problems by encouraging a diversified filter configuration. In this work, we apply MHE regularization to the 1D filters of the Wave-U-Net. We evaluated this approach for separating the vocal part from mixed music audio recordings on the MUSDB18 dataset. We found that adding MHE regularization to the loss function consistently improves singing voice separation, as measured in the Signal to Distortion Ratio on test recordings, leading to the current best time-domain system for singing voice extraction.

Automatic cover detection -- the task of finding in a audio dataset all covers of a query track -- has long been a challenging theoretical problem in MIR community. It also became a practical need for music composers societies requiring to detect automatically if an audio excerpt embeds musical content belonging to their catalog. In a recent work, we addressed this problem with a convolutional neural network mapping each track's dominant melody to an embedding vector, and trained to minimize cover pairs distance in the embeddings space, while maximizing it for non-covers. We showed in particular that training this model with enough works having five or more covers yields state-of-the-art results. This however does not reflect the realistic use case, where music catalogs typically contain works with zero or at most one or two covers. We thus introduce here a new test set incorporating these constraints, and propose two contributions to improve our model's accuracy under these stricter conditions: we replace dominant melody with multi-pitch representation as input data, and describe a novel prototypical triplet loss designed to improve covers clustering. We show that these changes improve results significantly for two concrete use cases, large dataset lookup and live songs identification.

Artificial intelligence is fast becoming an essential ingredient for an array of solutions across all industries – which promises to transform every aspect of our lives. While intelligence similar to (or perhaps even surpassing) that of humans may emerge in the not-too-distant future, it's clear from our collective everyday experience we are not there yet. We are, however, firmly on the path. And there are steps we can take to create the brighter future we desire. This is especially important when it comes to the media and publishing industry.

Google has never widely been considered the top banana when it comes to smartphones, a designation bestowed instead on Samsung or Apple, depending on whether your loyalties lie with Android or iOS. But the last couple of years, Google's Pixels have presented an awfully strong case: solid Android phones with superb cameras that you can usually get for less than you pay for a top Galaxy or iPhone. So it goes with the Pixel 4 I've been using over the several days. It has a 5.7-inch display and starts at $799 (or $899 for its larger 6.3-inch sibling, the Pixel 4 XL) and for the first time is being embraced by all the U.S. wireless carriers out of the gate; in the past years, Verizon had the exclusive. As with other Pixels, the obedient Google Assistant is readily at hand, summoned through a familiar "Hey, Google" or "OK, Google" command, tapping an icon, and now even by squeezing the sides of the phone.

In order to become and remain competitive, companies need to tap into the latest technologies and customer experience trends. Here are 6 customer experience trends every company must get ready for. It's no longer enough to have a surface-level understanding of customers. It's assumed every business should know who their customers are, where they live, how old they are, and other rudimentary info. Today, companies must go deeper, and with fast-advancing technology such as the Internet of Things (IoT), they are able to get a 360-degree view of customers and markets. Consider how Disney uses Magic Bands at its theme parks.

Artificial intelligence (AI) is already all around us in our daily lives.