Ahmadzei's picture
added 3 more tables for large emb model
5fa1a76
To use SpeechT5 with the pipeline, you'll need a speaker embedding.