The SSI263 codepaths are currently disabled. This means that any images that attempt to use this feature of the Mockingboard will not work correctly.
At a minimum the effort to support this feature would roughly be:
Find a good set of disk images that exercise the speech chips
Study SSI263 datasheets from ASIMOV
Augment our backend audio API (our shims to OpenAL, OpenSLES, etc) to support "one-shot" and repeating play of phonemes
Uncomment and refactor the dedicated SSI263 thread(s) to use pthreads
Possibly refactoring needed in our simplistic software mixer (in the OpenAL/OpenSLES/etc backend), to add support for more than just 2 voices (speaker and existing mockingboard)
Write new [tracing] tests that prove correctness and conformance of new code against a baseline
If newly enabled SSI263 codepaths add significant overhead for low-end Android/embedded devices, default-disable SSI263 through a user preference
The SSI263 codepaths are currently disabled. This means that any images that attempt to use this feature of the Mockingboard will not work correctly.
At a minimum the effort to support this feature would roughly be: