Ahmadzei's picture
added 3 more tables for large emb model
5fa1a76
The SpeechT5 framework consists of a shared encoder-decoder network and six modal-specific (speech/text) pre/post-nets.