This went viral just two days ago Suno challenger, “Music version of Sora”Officially revealed: official name Udiohas been completely free and open, and each person has enough per month 1200 first quotanetizens went crazy.
Someone actually wrote the first updateAltman releases GPT-5 quickly metal work.Ahem, more powerful ones even createdtalk show…!
Advertisement
The founder smacked his thigh after reading:This is definitely a feature we didn’t expect. The scene was as lively as when Sora was born. Not only does it have everything you need in various styles (classical, pop, rap, piano jazz, heavy metal, etc.), it also supports multiple languages, including Chinese and Korean songs.
You can enjoy the currentin the communitymost popular(76,000 plays) this lyrical songwith a total length of 3 minutes and 17 seconds, is quite beautiful.
Due to too many people playing, the server was crowded for a while (it’s ok now):
Advertisement
The quality is so high that even the famous player Spotify was directly cueed:
Buy it now while you can still afford it. (Manual dog head)
What exactly happened?
The musical version of Sora is here
As seen at the beginning, the “musical version of Sora” widely rumored by netizens is Udio. It has just been released as a Beta version, and the model behind it is v1 version, and it will be rapidly iterated in the future.
We enter the official website and complete the registration to use it for free: each person can create up to 1,200 songs per month.
The specific way to play is to enter text first, such as “a song about the experience of moving to a big city, heavy metal”.
Then the system will recommend some suggested tags to you, all about styles and genres, so you can choose whatever you want.As forlyricsIf so, you can write it yourself, or it can be a pure musical instrument, or let AI generate it for you.
After everything is set, click “Create” and wait for about 1-2 minutes before you get the result. (ps. Since there are still many people trying it now, the official explanation is that the generation speed is indeed slower than normal)
The result is two songs by default – the song titles have been chosen for you.
Click to check the results and view the lyrics.
If you are not satisfied, you can further edit, expand or remix it.
The extension can help us increase the default 33 seconds to 1 minute or even 3 minutes.
If you are satisfied with everything, you can click “publish” to publish it to the community.
I just clicked on a few pieces and they all really made me wow. This cheerful “Allegro Consort in C” must also be listened to by everyone: – With such an easy-to-use music generator, everyone is curious about how it is trained and what it is trained on.
In this regard, there is no official disclosure, but it does say that the product hasCopyright protection measuresthe model is equipped with automatic filters to ensure that the music created on the platform is original.
Built by former DeepMind engineers
According to reports, Udio was mainly built by a group of former Google DeepMind engineers.
Specifically, 4 of the 5 joint ventures are from Google DeepMind, namely:
David Ding
I have worked at DeepMind for more than 5 years. Before I left, I was the technical director of a team of 30 people (engaging in generative AI). Now I am the technical director of Udio. CEO.
Yaroslav Ganin
Originally from Russia, worked as an intern + full-time employee at DeepMind for more than 6 years, and is now a research scientist at Udio.
Charlie Nash
From the UK, I have been interning + working full-time at DeepMind for more than 5 years.
Conor Durkan
Also from the UK, he has served as a research scientist at DeepMind for more than 20 years.
The only person who is not a former DeepMind employee is Andrew Sanchez, who is the COO of Udio and graduated from Oxford University.
It is worth mentioning that Google DeepMind last NovemberA music generation model has just been released Lyria. When we compare Lyria's contributor list, we can find that the names of Udio's four co-creators are all listed:
Plus Udio happens to be toolast Novemberwas established, so some netizens “boldly speculated”: Udio is“Escape from Lyria” from Googlewhich they only spent 3 months retraining on their own machines.
In this regard, some netizens said, if this is really the case, should Google file a lawsuit?
However, we soon saw someone commenting that the underlying layers of Udio and Lyria are different.
We don't know exactly how yet. What is certain is that Udio, which has the “bloodline” of Google DeepMind, has indeed set off a carnival of AI music generation.
On Twitter, Grok even started a discussion on the topic “Is AI music a creative revolution or an industry threat?” The feeling is exactly the same as when AI painting became popular.
Such Udio naturally has high hopes from many big guys.
It is understood that in Udio’s current seed round of financing, investors include Co-founder and CTO of ins Mike Krieger,Head of Google Gemini Oriol Vinyals, and real artists— American rapper will.i.am(11.54 million Twitter followers) Wait.
What do you think of the effect of Udio? Which one do you prefer compared to Suno?
Reference links:
https://twitter.com/udiomusic/status/1778045340978933989
https://twitter.com/DavidDingAI/status/1778559157127254027
https://twitter.com/futuristflower/status/1778379253513351193