数据压缩1 | 浊音&清音&爆破音时域及频域特性

浊音&清音&爆破音时域及频域特性分析

概念区别

  1. 当气流通过声门时,如果声带的张力恰好使声带产生张弛震荡式振动,产生一股准周期脉冲七六,这一气流激励声道就产生浊音(Voiced Speech)或有声语音。
  2. 如果声带不振动,而声道在某处收缩,迫使气流以高速通过这一收缩部分而产生湍流,就产生清音(Unvoiced Speech)或摩擦音,或称无声语音。
  3. 如果声道在完全闭合的情况下突然释放就产生爆破音(Plosive Speech)。
    王炳熙,实用语音识别的奠基人
    [En]

    Wang Bingxi, the Foundation of practical speech recognition

简单地说,在语音学中,发音时声带振动的声音叫浊音,声带不振动的声音叫清音。辅音清晰,发音清晰,而大多数语言中的元音是发音的,鼻音、侧元音和半元音也是发音的。爆裂声是指发声器官发出的声音,在口腔中形成屏障,然后空气冲破屏障。

[En]

To put it simply, in phonetics, the sound in which the vocal cords vibrate during pronunciation is called voiced, and the sound in which the vocal cords do not vibrate is called unvoiced. Consonants are clear and voiced, while vowels in most languages are voiced, and nasal, side and semi-vowels are also voiced. The bursting sound refers to the sound made by the vocal organs that form a barrier in the mouth and then the air breaks through the barrier.

这里我利用 Audacity ,分别录制浊音 i ,清音 s 以及爆破音 b 进行时域频域分析,结果如下:

浊音时域及频域特性

数据压缩1 | 浊音&清音&爆破音时域及频域特性
数据压缩1 | 浊音&清音&爆破音时域及频域特性
结论1
在时间域上,浊音信号是短周期的,波形规则;在频域上,频率集中在低频区,共振峰(通道管的谐振频率)更明显,短期能量更高。
[En]

In the time domain, the voiced signal is short periodic and the waveform is regular; in the frequency domain, the frequency is concentrated in the low frequency region, the formant (resonant frequency of the channel tube) is more obvious, and the short-term energy is higher.

; 清音时域及频域特性

数据压缩1 | 浊音&清音&爆破音时域及频域特性
数据压缩1 | 浊音&清音&爆破音时域及频域特性
结论2
清晰的声音信号在时域上表现为低幅度和不规则的振动,在频域中分布在所有频段,总体波动相对平缓。
[En]

The clear sound signal shows low amplitude and irregular vibration in the time domain, and it is distributed in all frequency bands in the frequency domain, and the overall fluctuation is relatively gentle.

爆破音时域及频域特性

数据压缩1 | 浊音&清音&爆破音时域及频域特性
数据压缩1 | 浊音&清音&爆破音时域及频域特性
结论3
在突发声信号的时间域上,从规则(类似清音)到不规则(类似清音)有明显的过渡,从总体趋势来看,幅度下降;语音频率在频域更集中,但没有清音那样明显的共振峰,波动比清音更剧烈。
[En]

In the time domain of burst sound signal, there is an obvious transition from regular (similar to unvoiced sound) to irregular (similar to unvoiced sound), and from the overall trend, the amplitude decreases; the speech frequency is more concentrated in the frequency domain, but there is no formant as obvious as unvoiced sound, and the fluctuation is more intense than that of unvoiced sound.

; 总结

没想到CSDN第一篇博客献给了作业,第一篇小红书是为了打广告,第一篇公众号过稿是为了网红店……

可以肯定的是,他并不是没有外部力量(利益)就离开的。

[En]

It is sure enough that he does not go without external force (benefit).

这个小任务不涉及编程,暂时不会让人秃顶。感谢上学期的《数字音视频处理》和《百度百科》。

[En]

This small assignment does not involve programming and will not make people bald for the time being. I would like to thank * Digital Video and Audio processing * and * Baidu Encyclopedia * last semester.

Original: https://blog.csdn.net/qq_44100263/article/details/114451416
Author: 月婵婵
Title: 数据压缩1 | 浊音&清音&爆破音时域及频域特性

原创文章受到原创版权保护。转载请注明出处:https://www.johngo689.com/524892/

转载文章受原作者版权保护。转载请注明原作者出处!

(0)

大家都在看

亲爱的 Coder【最近整理,可免费获取】👉 最新必读书单  | 👏 面试题下载  | 🌎 免费的AI知识星球