Why are you charging for a beta program?
r
renderstorm1856
1st - why no signup for your feedback program using Discord, like every other ai company on the planet?
2nd why don't you interact with users in Discord like is traditionally done by ai generation companies? Much easier for everyone!!
3rd - why doesn't your platform follow prompts?
4th - why are there no user guides on Discord? Try visiting Midjourney on Discord to see how it is done!!!
5th - why don't you learn from the trials, errors, and successes of graphic ai, and incorporate similar solutions for music ai? Anyone remotely familiar with graphic ai will already be used to using references and combinations... if I want a female character in graphic ai (Midjourney , Stable Diffusion, etc.) I can prompt, "A 25 year old Jennifer Aniston|Taylor Swift, with blonde hair, pink t-shirt, blue jeans, walking by the beach" I could further reference camera equipment, lighting, etc. and those real-life reference will help influence the final image. I can reference a painting style, an artist, etc. And it doesn't mean I want to copy those references exactly!! It just gives me and ai common ground - a point of reference - so ai knows the right direction to go. If I add blonde girl from Argentina, that will be different than Taylor Swift from the USA. These are just references, pointers from which to work. It is insanely stupid that your platform doesn't allow for references. Anyone with any experience knows that references make it a hell of lot easier to get a skinny white blonde girl, than prompting every last detail of what you want. Similarly, references would be incredibly helpful in song creation. The lead guitar from Metallica AS A REFERENCE, or the raspy voice similar to Halestorm would make it so much easier to point ai in the right direction. It doesn't mean I want to copy those traits, it means use those as a reference.
A user on Discord advised, "If you don't get what you want the first time, try and try again" to which I responded, Well, that is certainly a financially beneficial business model for the company - hey our ai won't follow prompts like graphic ai does, and we can't really tell you how to get what you want, so just keep paying for credit after credit and you're bound to get something you like sooner or later, but we sure like that every time you try it costs you another credit. 😀 If I didn't have 11,000+ ai images under my belt, and was not familiar with how prompts should work, i would probably be more open minded about paying for this Udio beta program, but so far I am finding it difficult for them to justify billing users until they get the platform to work at least similar to ai graphics. I do see lots of chatter online about people complaining about how many times they must re-try to get Udio to produce something close to usable. Excuses like, music is newer than graphics ai, just don't resonate with an ai industry that is moving so fast, especially since the music side could easily have learned from the graphic history and be further along faster. But, yes, breathe, breathe, and if you get frustrated enough, the free market will provide a better solution... or you start your own company and do it better. So I get what you are saying. Doesn't make it right, but the thoughts are valuable. Thanks.
Udio can do better. You should. These loose ends that you presently have are unconscionable. Leaving users with no user guides, and guides that aren't updated each week as users find solutions and as your staff provides support ON DISCORD, puts your company and all users at a huge disadvantage.
Do better.
U
UdioAdam
Sorry for the frustration! A few thoughts on top of Shane's:
- We're using this third party system for garnering and replying to feedback because it's the best one we found.
- Many of us from Udio interact with our community extensively on Discord. This feedback system is in addition to that.
- Generative AI music is an extremely different beast from genAI imagery. Prompt adherence is a very tough problem.
- We recognize and regret that our documentation is poor. We're prioritizing improving it massively in Q1.
Sorin Acela
UdioAdam We've been seeing for months a group of web developers who cannot even design an ergonomic user interface. The interface is a terrible mess. Many settings have no effect. The Staff Picks are ridiculous. The remix of a small portion of a song is totally useless. You do need a guiding organizing mind or your progress will keep being slow.
Shane
To note, I'm not a full member of staff, so I can't answer all these questions for you.
- What do you mean by the feedback program? The TT testing? There was signup for it on the Discord, but it's no longer in an open period for new testers right now. It will open again at some point again for the next wave of testers when they want some more. I don't feel like there's a shortcoming in the current testers at all though, there are many and they are amongst the most experienced Udio members in the community.
- I can't really comment on this one. But I can note that the staff are all very busy.
- Such is the nature of GenAI, but particularly for music right now. It is a very difficult problem, but it's one that is always being worked on. Along with the output sound quality it is probably the highest priority for the devs, of course they want Udio to be able to follow prompts as well as possible, but it is a very difficult task. Here's hoping (and expecting) good improvement in this area in the future.
- This is planned. It is just a matter of resourses atm. Though that project won't be just yet, can put soon™ on that one.
- Music is a different medium to images and video and text. It comes with different challenges and obstacles, that take different forms to what you may not realize from the outside. But all is not as simple as it potentially seems. I am not a dev so I am not in a position to comment too much (and I wouldn't be able to anyway even if I were a dev). But things are being looked into in the future, some of the Udio devs literally come from DeepMind and Google, trust me they have intimiate knowledge and understanding of other AI platforms and ML techniques. The fact is (in my and many others opinion) that Udio is the best AI music gen tool in the world right now. And the fact that it is better in output sound quality than Suno, despite Suno having had more time and resources and money, should say something to the ability of the Udio devs and the difficulty overall in AI music gen. There's limited resources to go around and a very long road ahead, but things will inevitably keep improving over time.
Shane
0 - "Why are you charging for a beta program?" Because it costs a lot of money to process the generations and pay the devs / rest of staff. They can't keep giving away the product for free because they'd go bankrupt and nobody would invest in them. So it's either a case of charging for it, or ending it entirely.
It is up to users whether they consider the product worth paying for, despite it currently having some bugs and output issues. You can see that thousands of amazing songs have been created in Udio in it's current iteration. Nobody is forcing anyone to use or pay for the product if they don't want to. Many people still get great utility out of the product and love Udio, despite some of it's current short-comings. There's lots to learn on ways to potentially try and work around the existings issues, others are waiting for future updates to be able to complete some projects and only completing what ones that can do right now, some creating samples / tracks ready to go or edit in the future after certain updates.
It's a case of trying to adapt to what's available right now, and accepting where we are, realizing that progress takes time and trying to be patient but optimistic and looking forward to an even better Udio in the future. I personally consider it absolutely incredible and mind-blowing still that we have msuci technology like this at all. Though the human mind soon naturally gets used to, and has expectation of things, very quickly after they've become a main stay for a little while; I find it helpful to look back to where we were just 12 months ago in order to keep the perspctive well grounded.