It seems that they are using Nuance under the hood for the speech recognition.
You can see your currently installed language packs here:
RSSDK\contrib\Nuance\LanguagePacks
I am not sure if it would work, but maybe you could get the language pack from nuance and put them there.
German seems to be officially supported:
enum LanguageType {
LANGUAGE_US_ENGLISH = PXC_UID('e','n','U','S'), /** US English */
LANGUAGE_GB_ENGLISH = PXC_UID('e','n','G','B'), /** British English */
LANGUAGE_DE_GERMAN = PXC_UID('d','e','D','E'), /** German */
LANGUAGE_US_SPANISH = PXC_UID('e','s','U','S'), /** US Spanish */
LANGUAGE_LA_SPANISH = PXC_UID('e','s','L','A'), /** Latin American Spanish */
LANGUAGE_FR_FRENCH = PXC_UID('f','r','F','R'), /** French */
LANGUAGE_IT_ITALIAN = PXC_UID('i','t','I','T'), /** Italian */
LANGUAGE_JP_JAPANESE = PXC_UID('j','a','J','P'), /** Japanese */
LANGUAGE_CN_CHINESE = PXC_UID('z','h','C','N'), /** Simplified Chinese */
LANGUAGE_BR_PORTUGUESE = PXC_UID('p','t','B','R'), /** Portuguese */
LANGUAGE_RU_RUSSIAN = PXC_UID('r','u','R','U'), /** Russian */
};
