In January, Google unveiled MusicLM, an “experimental AI” tool capable of generating high-fidelity music from text prompts and hums.
The tool is now available to be tested by the public.
Google explains that at the public usage level, the tool works by typing in a prompt like “soulful jazz for a dinner party.”
The MusicLM template will then create two versions of the requested song for the person entering the prompt. You can then vote for your favorite, which Google says will “help improve the AI model.”
The model was trained on five million audio clips, or 280,000 hours of music at 24 kHz.
At the time of its unveiling in January, Google released a set of examples of the tool’s “audio generation” capabilities “from rich captions”, which you can listen here.
Google claims that “whether you are a professional musician or a beginner, MusicLM is an experimental tool that can help you express your creativity”.
The company released a ‘look behind the scenes yesterday at MusicLM used by a sound artist, a Google Arts & Culture artist-in-residence and a Google researcher:
Google also published an article in January describing the research conducted to develop the tool.
According to the Google researchers, “Future work could focus on speech generation, as well as improving text conditioning and voice quality. Another aspect is modeling high-level song structure like intro, verse, and chorus.
The research paper, which suggests that MusicLM, “further extends the set of tools that aid humans in creative musical tasks”, also added that “there are several risks associated with our model and use case. which he is attacking”.
According to the researchers, among these risks are that “the samples generated reflect biases present in the training data, which raises the question of the relevance of musical generation for cultures underrepresented in the training data, while raising concerns about the culture of appropriation”.
Another risk highlighted by the document was the “potential misappropriation of creative content”.
THE the researchers explained“In line with responsible model development practices, we have conducted an in-depth study of memorization, adapting and extending a methodology used in text-based LLMs, focusing on the semantic modeling step.”
“We strongly emphasize the need for additional future work to address these risks associated with music generation – we do not intend to release models at this time.”
Google MusicLM research paper
They said they “found that only a tiny fraction of examples were remembered exactly, whereas for 1% of the examples we could identify an approximate match.”
And then added: “We strongly emphasize the need for additional future work to address these risks associated with music generation – we do not intend to release any models at this time.”
“Seven years into our journey as an AI-driven company, we are at an exciting inflection point.”
Sundar Pichai, Google and Alphabet
Google’s surprise public release of MusicLM this week came on the same day that Google and Alphabet CEO Sundar Pichai announced a huge push into AI with a range of AI-powered updates for various Google products.
“Seven years into our journey as an AI-driven company, we’re at an exciting inflection point,” Pichai said in his keynote at the Google I/O 2023 event. Wednesday, May 10.
“We have the opportunity to make AI even more useful for people, for businesses, for communities, for everyone.”
As part of Google’s new AI push, the company is expansion its conversational AI tool and rival Chat GPT, Bard in over 180 countries after an initial launch in the UK and US.
Bard was also recently moved by Google to its “state-of-the-art language model”. PALM 2. Google says it’s “a much more capable large language model that features” advanced math and reasoning skills and coding capabilities“.
The public release of MusicLM comes at a time of growing unease around the use of generative AI in music.
One of the main reasons for industry concerns about the use of generative AI, which is trained on other music, is the risk of copyright infringement.
Last month, AI-generated music productions who imitate the voice of superstar artists dominated the headlines after a song titled Have the heart on the handwith AI-generated vocals copying the voices of Drake and The Weeknd, went viral.
The track, uploaded by an artist called ghostwriter, was later removed from the likes of Youtube, Spotify and other platforms. On YouTube, a confirmation of what triggered the track’s removal from that platform appeared on ghostwriter’s homepage. YouTube download now gone.
He read: “This video is no longer available due to a copyright claim by Universal Music Group.”
speaking on Universal Music GroupIt is Call for Q1 results last month, Sir Lucian GraingeCEO and President of Universal Music Group, noted that: “Unlike its predecessors, much of the latest generative AI technology [i.e. ‘fake Drake’] is formed on copyrighted material, which clearly violates the rights of artists and labels and will put the platforms completely at odds with partnerships with us and our artists and those who lead to success.
In his opening remarks to analysts on that same call, Sir Lucian Grainge also criticized the “content oversupply” that currently sees around 100,000 tracks per day distributed to music streaming services.
“Few realize that AI was once a major contributor to this oversupply of content,” Grainge said.
“Most of this AI content on DSPs comes from previous generation AI, a technology that is not trained on copyrighted intellectual property and produces very poor quality output with virtually no consumer appeal.”
The rise of AI platforms that allow users to create vast volumes of leads at the touch of a button has also revealed the potential for using generative AI to streaming fraud.
Earlier this month, the AI-powered music-making app Boomy, whose users created 14.4 million songs to date, said that Spotify had close its ability to upload songs to the DSP and that some previously uploaded tracks had been removed.
A Spotify spokesperson later confirmed to MBW that these “certain catalog versions” of Boomy had been removed because the streaming platform had detected the artificial airing of these tracks. (There was no suggestion that Boomy himself was involved in artificial streaming).
Boomy said on Saturday May 6 that “arranged delivery to Spotify news from Boomy artists has been reactivated,the company wrote on its Discord server on Saturday, May 6.
Although Spotify confirmed that it had made some tracks unavailable, it appeared that it was probably Boomy’s own distribution partner – Downtown– owned by DashGo – which had halted uploads to Spotify.
Only a small portion of Boomy’s tracks appeared to have been “grayed out” so that they could not be played. On Monday, May 8, there were no greyed out tracks on Boomy’s playlists on Spotify.The music industry around the world