We’re excited to announce our latest model update release , which harnesses the power of the Diffusion Transformer (DiT) after extensive research and testing. This version leverages DiT’s unique blend of diffusion models and transformer architecture to capture intricate musical nuances and complex temporal dependencies more effectively. By combining the iterative refinement process inherent in diffusion with the global self-attention of transformers, we generate music with enhanced coherence and expressive detail. Looking ahead, we’re eager to further evolve this technology and explore new creative possibilities in music generation.
These updates represent major strides in our ongoing efforts to create more responsive and musically intelligent virtual musicians.
Rhythmic Improvements
We have enhanced the synchronization between the virtual musicians and uploaded audio through a renewed rhythmic alignment structure. This update focuses on establishing and maintaining precise temporal relationships between musical components, ensuring that every note and beat is coordinated within the context.
Grid Alignment
The virtual musicians now have an advanced understanding of their position within each musical bar. Through this spatial awareness, the generated audio aligns with the project’s musical grid, creating a more precise performance that maintains timing throughout the song.
Musical Phrasing
Our new architecture demonstrates an improved capacity for understanding and generating natural musical timing. This results in more coherent and musically logical phrases, allowing virtual musicians to craft performances that flow naturally and respond musically to musical context, much like human musicians would.
Genre Compatibility
Through extensive model training across an expanded range of musical styles and performance scenarios, we have grown our compatibility to more genres. This comprehensive training enables our virtual musicians to authentically interpret and perform within various musical contexts, maintaining genre-appropriate characteristics while ensuring musical authenticity throughout different styles. As a reminder, all the data we use has been collected with consent, ethically.
New & Refreshed Track-Head
We’re excited to roll out an upgraded Track-Head that makes working with Aiode’s virtual musicians even more intuitive and efficient. This update is designed to streamline your creative workflow, giving you more control and flexibility directly from the Composition page.
With the new and refreshed Tracker, you can now:
- Mute and solo individual tracks to focus on specific elements of your composition.
- Adjust volume levels in real time to fine-tune the balance of your mix.
- Generate and download your music, without interrupting your creative process.
These enhancements are all about making it easier for you to shape and refine your sound.
Registered to the Beta and want to increase your chance of getting the coveted invite in our next batch?
Join our Discord Community and follow us on our socials.
Facebook | Instagram
Need help? Contact Us
Support@aiode.com