Clouds are not only a place for storing data, although they are perfect for this too. However, cloud technologies have other, less obvious to the uninitiated, use cases.
We’ll tell you some stories about how companies moved to the cloud and what came out of it.
First Example: Cloud Technologies, Neural Networks, And Music
The music industry is not only about creativity. After the vocals are recorded in a good studio, the drum tracks are prepared, deep and rich synth samples are cut, it’s time for the mixing and mastering engineers to work. These guys put together all the sound material, make it clean and of high quality from a technical point of view – so that the melody sounds readable and clear both on expensive speaker systems and in the cheapest headphones.
Let’s dwell on this in more detail. Mastering is a dull, routine, and repetitive process. An engineer plays a piece of music for several hours and tweaks the knobs for adjusting the audio frequencies so that the bass does not boom, the melody does not rip the ear, and the drums do not bite into the brain. As you might guess, this job takes a lot of patience, attention to detail, and experience. And a good specialist should work with music, an hour of work which costs a tangible amount of money.
And so, the owners of a small record company realized that could reduce the job of a mastering engineer of machine algorithms. Of course, these algorithms will not be simple, but they can still be described in the form of a program.
This is how a service appeared where programs analyze and mix music tracks using machine learning and neural networks. You download materials for processing, pay a few dollars for the program, and have tea. Towards the end of the cup, a link to the finished file appears, which (lo and behold!) It sounds like after the work of a top-notch mastering engineer; If a human mixed it, mastering would cost hundreds of dollars.
How does the service work? He leases cloud power, which neural networks are launched to process sound. It is challenging to buy special equipment for this yourself – it needs to be connected and configured. In addition, it is not very clear how much to take: if you buy a little iron, it will not be enough for all customers, orders will accumulate in the queue. And if you buy too much, some of it will be idle.
The option with cloud technologies for business is ideal because you can always get as many resources as you need and pay only for what is used in the cloud. A client has appeared – the company takes the required amount of capacity. Launches the neural network, receives the finished result and returns the power to the cloud provider. Four clients appeared at once – they took four times more capacity, used it and returned it. Budget savings, flexibility, and people don’t line up.