Generative AI tools for music artists

Generative AI can be useful for both a DIY music artists and an artist with a team. It really helps you to complete creative tasks using text, sound or pictures, which would take a lot of time manually. What a person does in tens of minutes, a AI can do in seconds – generative AI. Music artists at present prefer to use automation tools for related processes.

There can be a lot of way for use AI (as well as the services themselves based on neural networks). We’re talking about those web applications that are useful for musicians in regular content production.

Text AI: from searching any naming ideas to preparing a press release or posts on social networks.

Graphical AI: from creating cover art to searching for reference ideas for a photo shoot and corporate identity.

Sound AI: any tasks from cutting noise in audio to generating beats by click.

In general, there are a lot of such services – in this article we’ve collected in a selection those AI, that we tested in practice (ourselves or they were recommended by mates musicians).

Text AI

ChatGPT

ChatGPT is 100% the most famous AI service at present. This is a generative language model that generates high-quality coherent text and works with information on a given topic: from scratch or with reference material. And it is not officially available in Russia: to enter, you will need a VPN + registration through a foreign phone number. There are telegram bots through which you can access ChatGPT – they can be easily found by searching in telegram.

YandexGPT

Strong web service by Russian Яндекс development with free access. O lot of people use YandexGPT to search for rhymes and successful expressions, to prepare a content plan or blog posts. So if it’s hard for you to write about yourself on your own, feed the algorithms a list of facts about yourself and the upcoming release – after some attempt, you’ll get a draft press release that you won’t even be ashamed to finish writing yourself and send to editors.

GigaChat

The Russian language model from Sberbank, but so far more promising than fully trained. The creators position GigaChat as an assistant for marketers and copywriters, so in the future it will be possible to use it to proofread texts for posts, come up with a marketing strategy for a release or a line of merch products. You can access the beta version through a telegram bot.

Notion AI

One of the special features of Notion AI is the “app for everything”, which structure fans from large and small teams love to use. If you are not yet familiar with the application itself, we recommend testing it for personal tasks. And if you already actively use and store song lyrics, release strategies and even a content plan there, then an additional feature with artificial intelligence will help you write posts, create and control tasks for a release campaign, edit your own materials and suggest ideas for content and creativity. By the way, despite the English-language interface, Notion AI understands the Russian language well and works with it.

Graphical AI

Шедеврум

This is a neural network for generating images in text query format, created by Yandex in 2023. Шедеврум works in mobile applications on Android and iOS. You can download app from Google Play and the App Store. The neural network does not yet have an online browser version. The peculiarity of the neural network is that all images created in it end up in a common feed. You can see what neuroarts users have recently generated, subscribe to the authors, and follow the top images that have received the most reactions. Also you can download images from your feed, like them, or share them – for example, publish them on VK or send them to someone in Telegram.

Kandinsky

A neural network from Сбер that generates pictures based on text queries.
The main feature of the service is that it works with the Russian language and generates pictures that are close to photorealism. You can describe the image, set the style, and clarify details. Unlike most similar services, Kandinsky understands Russian-language queries quite well and takes them into account when generating them. You can access the service through the official telegram bot.

glif

This is a platform that combines several neural networks in one interface. There are ChatGPT, Stable Diffusion and Kandinsky. Among other things, on glif you can also combine neural networks with each other and come up with new mini-services within the platform. A new experimental feature of the platform called Controlnet Any Word is just such a combination: one neural network generates an image (for example, a room with cats), and the second carefully inserts the desired inscription of the required size right in the center of the picture, disguising it as the background. Ideal for those cases when there is no money for a designer, but you really want a cool art Easter egg with an inscription for your album.

Генератор обложек от ПАНЧ

Генератор обложек от ПАНЧ was created specifically for generating album covers and layouts for posts and stories on social networks. To create a cover, you need to configure the parameters: the genre of music, its mood, the style of the image, what you want to see on it. You can also specify the title of the track and the name of the artist (they will not be added to the picture). The generation result highly depends on the settings and luck: if you repeat the same request, the result will be the same every time

Midjourney

A neural network that draws images based on a text description in English from the user. It is enough to formulate a text query and the program will produce the result in the specified style. Midjourney can create paintings, make stylizations in the spirit of popular films, and also invent the appearance of fairy-tale creatures. Now Midjourney performed as a Discord bot. It’s convenient not only to generate pictures, but also to get involved in the life of the neural network community in general: participate in competitions, put together ideal queries, check the work of new versions.

Sound AI

Riffusion

The free neural network generates music based on a text request: write what you want to hear and press play button. Riffusion is very easy to use – you can combine non-standard requests, generate melodies non-stop and download short excerpts of what you get.

Mubert

Mubert creates music using samples from the sound library. Generation starts on the main page: you enter keywords, set the duration of the track, select the genre, mood and type of activity suitable for the music, wait – and get the track. You can regenerate the results you like – make a remix, as it is called on the site. But it will be possible to obtain the rights to the generated music only for money – and since right now the legislative framework for neural networks is not fully formed, it is better to take care of this in advance.

Soundraw

A neural network that creates interesting music based on keywords, genre and mood. Soundraw has a nice, intuitive interface and dozens of variations for every request. If desired, you can modify and change the versions you like. In the paid version of Soundraw, which costs $17 per month, you can download tracks with copyrights to them.

Melobytes

A popular project that provides free access with restrictions: the demo version offers only five requests per day. But you can sign up in Melobytes for several different subscriptions and get different types of access to the service, which gives a good selection of settings and good results.

AIVA

The project was developed in the format of generating music for advertising, video games and films. The neural network creates tracks from scratch and also helps to correct ready-made melodies. In AIVA you can create different variations of the same track. There is a free version – it already opens up some good opportunities (with the exception of the right to commercial use of the results), but all the functionality is available only for money. Flexible tariffs for subscriptions – you can choose the optimal one based on your interests and requests. The minimum cost is 11 euros per month.

Maroofy

A neural network that has analyzed more than 120 million tracks from the Apple Music catalog and can make good recommendations. If you enter the artist and track name in the search, Maroofy will evaluate its rhythm, tonality and other parameters, and then create a playlist of similar tracks. The interface is very simple and works with tracks all over the world.