add audio_data to buf when speeking is False for non-stream inferring

weiliang

2023-03-24 1b8419d83267954b861a86f5ba1ccc6f60fe9098

1
2
3
4
5
6
beam_size: 10
penalty: 0.0
maxlenratio: 0.0
minlenratio: 0.0
ctc_weight: 1.0
lm_weight: 0.7