Why and how the text could be Bavarian

[JoJo_Jost]

But still, if it’s a cipher, the decryption should be unambiguous. A well-known example of a homophonic cipher is Copiale — this essentially confirms it. After removing the Latin letters, it instantly succumbed to cracking, given its complexity.

The key lies in the structure:

If there are few homophones and clear word boundaries, it’s usually relatively easy to decipher.

If there are many homophones with a well-balanced frequency, it’s already much more difficult.

But if there are also variable spellings, zero characters, obscured word boundaries, or context-dependent selection of homophones, it becomes considerably more difficult.

If the choice of homophone itself depends on position, word length, or neighboring characters, strictly speaking, it is already more than a simple homophonic substitution—and it can become extremely difficult.

This is especially true when the underlying language is subject to highly flexible spelling conventions, as was the case in Middle High German.

VBM breaks down word boundaries and operates at the level of consonant clusters and vowel bridges—that is, not at the token level. If both are interpreted as homophones, it becomes nearly impossible...

Could you clarify whether the cipher provides different translation options for a single string (for example, two meaningful translations), or whether all translations except one are meaningless?

I've done a lot of calculations and analysis, but without plain text, it's hardly realistic to answer such questions. However, it seems that, when it comes to vowels, multiple bridges represent a single vowel, and the most common consonant clusters can be represented by several different VMS Cluster.
In addition, there’s a second level—besides the families—and that seems to have been encoded differently yet again.

But, to be honest, this is all just speculation, of course.