Thanks for getting in touch! OK, so, I’ve taken some time to look at your logging output and I think what I’m seeing is that this isn’t using one of the two supported 8khz acoustic models that ship with RapidEars, is that correct? It looks like this is probably one of the 16khz Voxforge models if I’m not mistaken, which is why the performance is going to be slow with RapidEars. Can you let me know if there are any other modifications before we look into if there are any ways to improve the performance for the slower 16khz models on the old devices? A few of the other Sphinx runtime arguments look a bit unexpected to me so it would be good to know if any other changes have been made so that the troubleshooting time can be put to the most efficient use.