Spaces:
Running
Introducing Version 3 - Studio-Quality Mastering with Enhanced Clarity and Faster Processing
Hello Everyone,
I'm thrilled to share with you the results of our effort in advancing our experimental software. Our journey has been marked by over 700+ mastering sessions in the last several weeks, a testament to the dedication and enthusiasm of this project.
The feedback on the audio mastering engine's performance, quality, and user-friendliness has been crucial. Your insights have significantly contributed to refining the neural network architecture and enhancing the overall mastering experience.
Introducing Version 3 - Clarity, Precision, and Efficiency
We are proud to present the latest model (v3), featuring remarkable improvements in clarity and a more natural sound profile. This version stands out from its predecessor (v2), overcoming the "grainy" output issue, and reflects our dedication to making professional mastering accessible to artists everywhere.
Our tests confirm v3's durability and stability, showing a significant reduction in static and ringing issues encountered with certain tracks in previous versions. The model's performance and user experience have been thoroughly improved, ensuring efficient render times and reliable results for our users.
Current Status and What to Expect
The new model is now available by default, marking a significant milestone in our project's timeline. We've managed to introduce these enhancements ahead of our expected timeframe, thanks to steady progress and feedback from others.
Your support has been invaluable, and I'm thrilled to share this enhanced mastering experience with you.
Sincerely,
Reed Mayhew
Disclaimer: We are excited about our developments and uphold the integrity of our updates. However, please understand that the information provided is based on current progress and not a binding commitment for future releases. Our AI mastering services are currently available free of charge, given that hosting expenses remain within manageable limits. This condition is subject to change at any time for any reason, such as increased utilization of our platform, and more. Our goal is to keep you well-informed, actively engaged in our journey, and enjoying our offerings, recognizing the evolving nature of experimental technology and the fluctuating demand for online music platforms.
I think Version 5 is broken. Instrumental and "Regenerate High Frequencies" together just "regenerates" the "high frequencies" and nothing else.
*on the website i mean
Hey, Kaiden!
I was able to check the code and verify it is functioning properly. Here's what happened:
If you make a mastering request with "instrumentals" checked and the render comes back without the vocals removed, it means our main server was temporarily offline. Your request went to our backup server, which doesn’t do stem separation before mastering since it runs an optimized version of our model on a CPU. This also applies to advanced features like regeneration.
Usually, waiting 3-5 minutes and trying again will give the backup server enough time to boot up and process your request fully.
I'm working on adding a message box that will notify users if an error occurred and explain what went wrong. In the future, a message will display this information. Sorry for the confusion!
@Kaiden423 I've added a server response box that will appear below the audio output to inform you if an error occurred during the render request. This should help avoid any confusion if the render you receive is missing some of the changes you requested.
