The most recent update has caused significant disappointment for me, as it has for many other users. It is extremely frustrating that since this update, it has become difficult if not impossible to create high quality music. Features that previously worked reliably are now producing noticeably weaker or unusable results.
This frustration is clearly reflected throughout the community, where a large number of users are openly expressing their dissatisfaction. This raises serious concerns: How far is this decline expected to go? And more importantly, what is the future vision for Tunee?
As users, we expect transparency, clear communication, and a well-defined roadmap for the platform’s development. Tunee once showed great potential and served as a powerful creative tool for many. It is therefore particularly disappointing to see that trust has been noticeably undermined by this update.
I sincerely hope this feedback is taken seriously and that meaningful improvements will be implemented soon to restore the platform’s original quality standards.
@von_corso I understand this very well, but I think Tunee will try to make some changes to keep users happy and avoid losing them. That would be a real shame. I have tried a few AI tools, but I was quite happy with what Tunee offers. I really enjoyed making music on Tunee and would not like to leave the platform. Let’s see how things develop, I will wait a little longer.
I’m a user and today Tunee seems to be treating styles as loose tags, not as real musical structures. That is, it recognizes “Bossa Nova” as: medium tempo + chords with 7th/9th + light drums. But it completely ignores the guitar pattern, the rhythmic cell, the bass/guitar relationship, which are the soul of the style. I realize it “doesn’t know what bossa nova rhythm is”—because, in fact, the current model isn’t respecting that. When this happens, the generic prompt stops working. Phrases like “traditional Bossa Nova,” “Brazilian Bossa Nova,” “smooth groove” are no longer enough. The model needs to be structurally forced, describing function by function. Even then, in some cases, it doesn’t solve the problem, because the issue is in the engine, not the prompt text. I really enjoyed producing music on Tunee and wouldn’t want to leave the platform.
I’m familiar with these kinds of issues as well, so it’s not really fun anymore. As mentioned above, however, I hope Tunne does something about it so that users don’t leave Tunee.”
Hello, thank you very much for taking the time to share such detailed and thoughtful feedback. We truly appreciate the depth of your explanation and your long-term engagement with Tunee.
You’re absolutely right in pointing out that, at the moment, some style descriptors (such as Bossa Nova) may be interpreted more as high-level style tags rather than fully enforced musical structures. In certain cases, this can lead to results that capture surface elements (tempo, harmony, general texture) but fall short in reproducing deeper stylistic characteristics like rhythmic cells, guitar–bass interplay, or genre-specific groove patterns. This is a known limitation of the current generation engine, and we understand how frustrating it can be when prompt refinements no longer bring the expected improvements.
From a technical perspective, this issue is less about prompt wording and more about model capability and enforcement at the engine level. The team is actively evaluating these gaps, especially for structurally rich genres such as Brazilian styles, jazz subgenres, and rhythm-driven music.
As a temporary workaround, we recommend trying the Mureka O2 model, which in some cases provides more stable results for genre coherence and instrumental relationships. While this won’t fully solve all structural issues, many users have reported slightly improved consistency when working with complex styles.
We completely understand your disappointment with the recent update and the concerns about overall direction and quality. Your feedback reflects sentiments we’re seeing across the community, and it has already been shared internally with both the product and model teams. Improving musical structure awareness and restoring reliable creative control are key topics currently under discussion.
Thank you again for your honesty and for sticking with Tunee. We genuinely value users like you, and we hope future improvements will better live up to the creative standards you’re looking for. If you’re open to it, any specific examples or project links would also be very helpful for our technical review.
Are you working on anything like suno’s personas mode? Being able to create a song and reuse the vocalist from said song again. I would like to create full bands and albums using the same vocalist and style in order to keep continuity.
What happens when a rap song you put a lot of heart and soul into gets turned into a melodrama? You love rap, but the channel turns it into a melodrama. Yes, if it could be the same, you could just upload a beat, create a similar track, and rap over it, doing exactly what you requested. But since January 1st, things have gotten progressively worse. When you upload a rap song, the channel turns it into a melodrama, and when you ask them to re-upload it, they say, “Sorry, I can’t rap, I can only sing or compose a beat.” You upload a beat or an instrumental track, and it shows up with a duration, but it’s buggy and keeps malfunctioning. You have to redo it many times, and you’re very frustrated, so you switch to another website.
@ho_ng_nguy_n I can clearly understand this problem as I have had similar experiences. I am always faced with the problem that rap content is not implemented as intended. When attempting to create rap, the output is automatically converted to a sung ballad or melodramatic track, even though rap is clearly stated as the desired style.
When a lot of time, dedication and creative effort is put into a rap song, it is extremely frustrating when the end result is stylistically altered and no longer reflects the original intent.
Additionally, technical issues with instrumentals or beats regularly occur, including incorrect track durations, playback errors, and inconsistent results that require repeated attempts. This leads to a significant loss of time, motivation and resources. For this reason, I have largely stopped uploading content, as repeated attempts rarely produce satisfactory or high-quality results. Furthermore, these repeated unsuccessful attempts inevitably use up available credits without producing any useful result.