Improving Fluency of Spoken Mandarin for Nonnative Speakers by Prosodic Boundary Prediction Based on Deep Learning

Table 2

Experimental results of PW and PPH boundary prediction.

Model	Precision (%)		Recall (%)		score (%)
	PW	PPH	PW	PPH	PW	PPH
BLSTM	94.63	87.84	97.38	74.76	95.98	80.77
seq2seq	95.54	83.40	97.77	81.76	96.64	82.57
seq2seq+attention	97.50	84.21	98.78	81.59	98.14	82.88