Google created an artificial intelligence capable of making songs based on text descriptions. The project is called MusicLM and is “a high-fidelity model generator of music from text descriptions”. MusicLM was trained on more than 280,000 hours of music content to learn how to generate coherent melodies for each context.
Looks like Google has woken up!
After the blows it received from OpenAI and its AI models, Google decided to enter the fray with force.
#BigTech has been investing in AI for a long time and often impressed the world with its models, but the difference was that OpenAI was much more adept at making these technologies easier to use. After the crash, it looks like they’re coming back to the race and releasing Bard which is still in its experimental phase.
Now, an impressive new “generative AI” system from Google can generate music in any genre with a text description. It’s detailed in a scholarly article, “MusicLM: Generating Music From Text,” which is worth a read. This is part of a broader trend in AI music creation, which includes innovations like the AI song cover generator, tools designed to create unique renditions of existing tracks.
Named MusicLM, it is not the first generative AI system for music. We’ve already seen other attempts, including Riffusion, an AI that composes music by visualizing it, as well as Dance Diffusion, Google’s own AudioML, and OpenAI’s Jukebox. But due to technical limitations and limited training data, none have been able to produce songs that are particularly complex in composition or high fidelity. MusicLM is perhaps the first that can.
OpenAI (which is the currently “apple of the eye” of the AI universe), released an AI music generator called Jukebox in 2020, but it did not receive the same level of adoption as its text and image counterparts. Although Google itself released AudioLM in the previous year, it did not receive much attention from the public either.
MusicLM is different because, according to its developers, it can create a “long and coherent song at 24 kHz” through extremely complex descriptions. Google has already released some of the songs created by MusicLM on GitHub, along with the instructions that were given to the AI. However, there is already controversy surrounding AI-generated art, with MusicLM’s developers acknowledging that at least 1% of the songs created by the model were plagiarized, making it difficult for Google to release it to the public.
Of course, MusicLM isn’t perfect, it’s far from it. Some of the samples have a warped quality to them, an inevitable side effect of the training process. And while MusicLM can technically generate vocals, including choral harmonies, they leave a lot to be desired. Most of the “lyrics” range from broken English to pure gibberish, sung by synthesized voices that sound like amalgamations of various artists. But it is another step in the evolution of “generative AI”.
Despite these concerns, Google’s move to develop MusicLM comes at a time when people have begun to question the company’s future after the release of the latest version of ChatGPT. Some are wondering if MusicLM is Google’s response to these concerns. However, the model may not be released to the public due to the fear of cultural misappropriation, plagiarism, or misrepresentation.
The advancement of AI and other technologies is inevitable, and our role is to understand and seek to benefit from these movements. Google’s MusicLM is a clear example of how AI can be utilized in creative fields, and we can only imagine the possibilities that lie ahead. The future of MusicLM is still uncertain, but it is sure to shake up the music industry and force us to rethink what we know about music creation.
Other AI-generated Music Tools:
There are many AI-generated music creation tools available in the market, and here we have covered a few of them, we may create a separate article listing all the tools that we can find and update the link here.
Creating Spectacular Results: Riffusion’s Journey
Riffusion, a music generation project created by roboticist Heik Martiros and software developer Seth Forsgren, has shown that AI-generated music is not only a possibility but also an impressive reality. Using Stable Diffusion, an open-source AI program, they were able to train it on spectrogram images combined with text cues to generate music.
How Riffusion Works
Riffusion’s process involves passing the music through AI imagers to create spectrograms, which are then converted into audio clips. The team at Riffusion trained Stable Diffusion to generate spectrogram images accurately enough to convert them to audio clips. Visitors to the website can add their own text prompts, and Riffusion will generate a spectrogram that can be played on the site as an audio clip.
Riffusion’s Impressive Results
The quality of the music generated by Riffusion may not be of the same caliber as man-made music, but it is certainly not as bad as one might expect. Riffusion has been able to generate music that includes Eminem and Kpop style rap, but the function of generating letters is not yet perfect. Instead of text, listeners will hear melodious human gibberish that still matches the tone of the song.
The success of Riffusion has opened up new possibilities for AI-generated music. As technology continues to advance, it’s only a matter of time before AI-generated music becomes a regular part of the music industry. While it may not completely replace human-made music, it will certainly provide musicians with new tools to explore and innovate.
With Mubert AI, users can choose the style, genre, and mood of their music, as well as the length of the track. The best part? It’s all copyright-free unless you need to create a large number of songs and tracks.
Mubert AI was created using over 4,000 samples of real musicians. The AI technology uses these samples to create original music tracks that are tailored to your specifications. The result is a completely unique sound that is copyright-free and can be used for any purpose.
Ready-Made Tracks and Collaboration with Musicians
In addition to creating your own tracks, Mubert AI also offers a library of ready-made tracks that have been created by other users. These tracks are available for free and can be used for any purpose. Moreover, Mubert AI is committed to supporting musicians who collaborate with their tool. The company offers a percentage of its revenue to these artists, ensuring that they receive compensation for their work.
Testing Mubert AI
If you want to test Mubert AI for yourself, all you need to do is visit their website and make your selections. The tool will generate a unique instrumental soundtrack that is tailored to your preferences. With Mubert AI, you can experiment with different styles and moods to find the perfect soundtrack for your content.
AI Sound Tools: New Tools That Can Enhance Your Audio Experience
Here are some of the latest AI sound tools available for users:
FreddieMeter: Channel Your Inner Freddie Mercury
If you’ve ever wondered how close you can get to Freddie Mercury’s singing style, then FreddieMeter is perfect for you. This tool, developed by Queen and YouTube, uses AI to analyze your singing voice and compare it to Freddie’s iconic voice. It rates your performance on a scale of 0 to 100, providing feedback on your pitch, melody, and timbre. FreddieMeter is a fun way to test your singing skills and learn from one of the greatest rock legends of all time.
Soundraw: AI Music Generator
Soundraw can create multiple pieces of incidental music in seconds. Choose a mood, genre and theme and the AI system creates 15 short music clips for you to choose from. Whether you’re a content creator or a musician looking for inspiration, Soundraw can help you generate unique and original music.
Murf: Create AI Voiceovers
Murf is a sophisticated AI service that can create natural-sounding speech in 120 voices and 20 languages. You simply input the text, and Murf generates a voiceover that sounds like a human. The latest feature of Murf is Voice Cloning, where the service attempts to clone human voices, which can be useful for creating voiceovers for various applications.
NaturalReader: Read Text Aloud
NaturalReader is a voiceover service that reads text aloud in several hundred available voices. It takes a picture of your smartphone, extracts the text, and reads it aloud, making it ideal for individuals who need to listen to their emails, articles, or any other written material while on the go. The Personal Reader app can also read eBooks, PDFs, screenshots, and many other text sources.
Cleanvoice.ai: Clean Up Audio Recordings
Recording audio can be a challenging task, especially when there are unwanted background noises, umms, ahhs, and lip smacks. Cleanvoice.ai is an AI tool that helps you clean up your audio recordings by removing unwanted noises and leaving only the essential audio. It works with accents and multiple languages, making it ideal for individuals who record audio in different languages.
AI-Generated Fake Podcasts: An Interesting Experiment
Eric Borgos, a software developer, created an AI-generated fake podcast to show how AI can be used to create fake content. Even the host’s photo is fake, generated by the fake people website Bored Humans. This experiment shows the potential of AI in creating fake content, which is a significant concern for the audio industry.
The Rise of AI-generated Content: Can Machines Replace Creativity?
The emergence of artificial intelligence has revolutionized various industries, including advertising and marketing. With the ability to generate images through text input, ChatGPT has proven to be a powerful tool in the creation of effective marketing texts. However, as the world becomes increasingly reliant on big data and relevance algorithms, the importance of creating content that is both creative and relevant has become a critical challenge.
For over 15 years, journalists have struggled to create headlines that capture the essence of their articles while also appealing to search engines. In the digital age, relevance is everything, and without it, even the most exceptional content can easily go unnoticed. With the vast amounts of data available online, the web has become an entity that knows everything about us, from what we want and what we consume to our likes and dislikes. This information is used to personalize our online experiences, presenting us with content that is tailored to our preferences.
In terms of creativity, something is still missing in AI. The “soul” perhaps?
Despite its promising potential, AI-generated content still lacks the depth and soul that comes from human creativity. While machines have shown impressive results in fields such as law and interpretation, where rules and laws can be applied, the ability to create something from scratch is still missing. Even in areas such as poetry and music, where AI has made significant strides, the results are often rough and generic. The creative work of AI, in many ways, still remains in its infancy.
One possible explanation for this is that creativity, by its very nature, is difficult to quantify and measure. It is an expression of the human soul, and as such, is often unpredictable and subject to change. The best creative works are often a reflection of the creator’s emotions, experiences, and unique perspectives.
The challenge for AI is to replicate this level of creativity while also providing relevant and engaging content. While systems such as Google’s MusicLM have shown impressive results, the question remains whether machines will ever be able to fully replace the creative genius of human beings.
Perhaps brilliant artist Nick Cave has a plausible explanation as to why the creative work of AIs is still primary, visit the below link to read his views on this topic:https://www.theredhandfiles.com/chat-gpt-what-do-you-think/
