Should artists be worried about Melodio, the new generative AI music streaming platform?The day we feared but expected is here. Chinese firm Kunlun Tech says it’s released the “first-ever” music streaming platform for generative AI music. Request a genre, mood, or any other prompt you want, and hear Melodio play non-stop music in the style you’ve chosen.
OK, before we freak out, let’s see what’s up.
Melodio is technically not the first AI music streaming service. Aimi has been around since 2021 and generates music based on material provided by established artists, such as Roska, Black Loops and Shanti Celeste. A similar platform is Endel; specific to mindfulness, focus and sleep, artists including James Blake, Richie Hawtin and Grimes have provided material for the generative soundscape platform.
The key difference with Melodio is the intention and the material. Kunlun Tech says that the platform “revolutionises music streaming” with its prospect of personalised, prompt-focused music. The examples it provides in a press statement include ‘energetic music for a long drive’ and ‘mellow tunes for morning coffee’. You can also change the input prompt at any time during the stream to change the style of the music. As the stream plays, you can save and share your favourite ‘music’ or segments of audio, which presumably impacts the rest of the stream.
But where does this music come from? Aimi and Endel rely on artists and creators to provide material — what’s Melodio’s training data?
It’s called SkyMusic 2.0. Kunlun says in the press release that this engine “leverages LLM technology to achieve an integrated end-to-end music generation process” that “boasts the capability to produce six-minute stereo songs with a 44,100Hz sampling rate and supports lyrics input exceeding 500 words.”
Alright, so no word on where the material is sourced.
We reached out to a Kunlun Tech representative for more information. The company said: “Different from the mainstream path of the industry, SkyMusic adopts the route of self-developed large-scale model music audio generation technology. This route directly realises the integrated end-to-end music generation of instruments, vocals, melodies, volumes, and notes through large model technology, which is extremely difficult technically, and only a handful of top players in the world, including Kunlun, participate.”
Still kinda mysterious — we’ve asked for some further info. If the company (which owns web browser Opera and formerly owned dating app Grindr) is operating similarly to GenAI platforms Suno and Udio, it might be about to face a legal firestorm. In June, the RIAA filed a lawsuit against the two companies for alleged copyright infringement en masse, seeking damages of up to $1.5 trillion from Suno and $1.36 trillion from Udio. Yikes.
Anyway, another crucial question is: Does Melodio actually sound convincing enough to replace anything in your music library? Not exactly. You can’t generate music that’s absent of lyrics, which are often corny and sung by robot-esque vocals, and the instrumentation is generic and ripe with artefacts. It’s not fit for leisurely listening.
However, its potential might not lie in replacing your favourite music, but instead as a supplement for background music. Think coffee shops, in-game music for video games, and other incidental music moments. As Barack Obama said last year, elevator music may be the first to be replaced.
Kunlun adds that, per Statista, “the global music streaming industry reached $19.3 billion in 2023. On Spotify, nearly 50 per cent of users have listened to playlists with mood tags, indicating a strong demand for scenario-oriented music. Compared to traditional music platforms, AI music streaming can tailor exclusive music to each different scenario, redefining the form of music and enabling users to enjoy customised musical companionship anytime, anywhere.”
Melodio probably won’t catch on in the way you’d expect it to, but it may become a powerful tool for certain markets and creators.
There’s more: Kunlun has also announced Mureka, which — similar to Suno and Udio — lets you generate music with a prompt. You can then download the track using Credits on your account; you’ll start with six but can purchase more to download more songs. You can even sell your AI songs on the Mureka Store — Kunlun claims this lets you “explore new business models for AIGC”.
What’s vital here, for everyone’s sake, is that Kunlun is playing fair from a licensing perspective. There’s no doubt that the industry at large will be keeping an eye on Melodio and Mureka to ensure no artists are being exploited.
If you’re intrigued, you can try a demo on the Melodio website. Just type in a prompt and hit that Brat green button stamped with “Let’s Vibe.”
To keep up with new developments in music tech, sign up to our newsletter. 
The post Should artists be worried about Melodio, the new generative AI music streaming platform? appeared first on MusicTech.