Home › Forums › OpenEars › feat.params › Reply To: feat.params
feat.params are the last read, so they will end up used even despite the hardwired values in OpenEars to the best of my recollection. Keep in mind that there are reasons I haven’t included various other languages (reasons include but aren’t limited to accuracy, conflicts with required OpenEars settings, resource requirements which don’t fit with mobile, and licenses of some models) so if you make your own, I will probably not be able to offer any support with it, including their use with paid plugins. Not because I don’t want to but just because they aren’t part of my testbed and the scope of OpenEars and its support is limited to the models I ship and their shipped configuration settings (such as feat.params) because I can include them in testing which has reasonably limited constraints.
For this reason I always recommend first setting up an Ubuntu VM with a relatively recent version of Sphinxbase and Pocketsphinx and getting everything working with your new acoustic model and language models, and then trying to bring the new model into OpenEars after it is 100% known-working in standard Sphinx, rather than first working through any issues in OpenEars and trying to troubleshoot it here.