[1]陈永彬.汉语连续语言的自动分段[J].东南大学学报(自然科学版),1987,17(4):8-14.[doi:10.3969/j.issn.1001-0505.1987.04.002]
 Chen Yongbin(Department of Radio Engineering).Automatic Segmentation of Chinese Continuous Speech[J].Journal of Southeast University (Natural Science Edition),1987,17(4):8-14.[doi:10.3969/j.issn.1001-0505.1987.04.002]
点击复制

汉语连续语言的自动分段()
分享到:

《东南大学学报(自然科学版)》[ISSN:1001-0505/CN:32-1178/N]

卷:
17
期数:
1987年第4期
页码:
8-14
栏目:
本刊信息
出版日期:
1987-07-20

文章信息/Info

Title:
Automatic Segmentation of Chinese Continuous Speech
作者:
陈永彬
南京工学院无线电工程系
Author(s):
Chen Yongbin(Department of Radio Engineering)
关键词:
声母 调母 帧平均能量 帧平均跨零数 自动分段
Keywords:
initial final frame average energy frame average zero-crossing number automatic segmentation
分类号:
+
DOI:
10.3969/j.issn.1001-0505.1987.04.002
摘要:
汉语是由单音节字组成的,而每个音节字又是由声母和调母组成的,分段就是从语言流背景噪声中检取出各音节字并分割出各音节字中声母和调母的范围。本文根据汉语时域参数的统计分布,设置若干阈值,用自动分段程序得出寂静、声母和调母的范围;以语言信号流波形的人工分段结果为标准,得出两句汉语(一句为女声,16个音节字;另一句为男声,9个音节字)的自动分段结果,其准确率达97%以上。
Abstract:
The Chinese continuous speech is composed of monosyllable words, while monosyllable is composed of initial and final. The segmentation technique aims at detecting the monosyllable word from background noise and classifying the initial and final of monosyllable. According to statistical value of time-domain parameters of Chinese speech, this paper presents an auto-segmentation program and sets several thresholds to get the areas of silence, initial and final of syllables. In experiment, two Chinese speech utterances (female: 16 charactors, male: 9 charactors) are segmented. It shows that auto-segmentation accuracy as compared with artificial segmentation is up to 97%.
更新日期/Last Update: 2013-05-01