We are building an autonomous home robot with NLP capabilities. SLAM part of the equation only requires CPU (OpenVSLAM). We are in between choosing TX2 NX or Xavier NX. We are wondering if Xavier can handle high quality ASR and TTS? What about TX2 NX?
The benchmarks are for vision DNNs but I am more interested in NLP, specifically in ASR (Automatic Speech Recognition) and TTS (Text To Speech). I can see the application of BERT in vision works on VOLTA or higher, but I have no idea how applicable it is on BERT’s application on NLP. Do you have performance figures for NLP for Jetson?