top of page

BRAPA BETA: Experimental Models in Development

  • Writer: Team BRAPA
    Team BRAPA
  • Jun 27
  • 1 min read

Updated: Jun 28

Team BRAPA is committed to the continuous evolution of its products and, as part of this process, we present BRAPA BETA - a line dedicated to the development and improvement of vocal synthesis technologies.


Objective of BRAPA BETA

The BRAPA BETA models are experimental versions created to test new approaches:

  • Voice model training

  • Labeling techniques

  • Improved support for cross language

  • Vocal modes and recording methods

The main purpose of these models is to collect data and technical feedback from the community in order to optimize the quality and performance of official releases.


Differences from Official Releases

As this is a line in the testing phase, the BETA models are considered independent projects from Team BRAPA's main line. The features implemented in these versions may not be incorporated into the final products.

To facilitate identification, BETA models have:

  • Icon with the silhouette of the original model

  • Specific download location

  • Symbol “β” following the model name (e.g. Umidaji 110β)

In addition to voices for DiffSinger, they may also be available in the future:

  • Experimental phonemizers

  • Tools and applications under development


An image showing the difference between Release and Beta

Feedback Request

As the BETA models are in the process of being adjusted, it is possible that they may present inconsistencies or errors. We therefore encourage technical users and enthusiasts to test and share their analysis via our feedback form. Your contribution is fundamental to the refinement of these technologies.

Thank you for your support and interest in our work.


Team BRAPA Brazilian in Synthesis

Note: BRAPA BETA models are intended for testing and research purposes. For use in final projects, we recommend waiting for the official launches.

bottom of page