Hi, I am writing to provide feedback on the recent changes to the audio upload process. Previously, when I uploaded a track, I could easily cover it while keeping the original vocal melody and pitch intact. However, since the addition of the three new options (Style Reference, Music Extension, and Music Analysis), the results have significantly degraded.
Even when using “Style Reference,” the AI completely changes the vocal melody, adds an unwanted accent to the singer, and loses the original song’s structure. None of the current options allow me to simply change the music style/arrangement while strictly preserving the uploaded vocal’s original melody and timing as it did before.
The current system feels like a step backward for users who want to create high-quality covers of their own recordings. Please consider restoring the previous functionality or improving the “Style Reference” to be more faithful to the original input.
Best regards.