3

I've come across a couple of vowel diagrams. Here one:

enter image description here

Here's another:

enter image description here

I'm not really sure how to interpenetrate these charts. Any ideas?

6

In pronouncing any given vowel, the recorded frequency spectrum will show distinct amplitude peaks (of the frequency spectrum). These peaks are called formants, and it is commonly accepted that the two lowest frequency formants taken together are enough to characterise the vowel.

The first formant F1 is the lowest frequency formant, and the second formant F2 is the second lowest frequency formant, etc.

enter image description here

Spectrogram of American English vowels [i, u, ɑ] showing the formants F1 and F2. The darkness of the band corresponds to the height of the amplitude/frequency plot; darkest areas show local peaks. From https://en.wikipedia.org/wiki/Formant#/media/File:Spectrogram_-iua-.png

The F1 frequency (range) corresponds to the perceived openness/height of the vowel, and the F2 frequency (range) corresponds to the perceived backness of the vowel.

enter image description here

Wikipedia IPA vowel chart

To demonstrate the effect of the F1 and F2 components on vowels, you can use artificial vowel generators which take these parameters to produce audible vowels, e.g. http://auditoryneuroscience.com/vocalizations-speech/two-formants.

As an example, consider the

  • Chengdu /o/ sound, which the graph says is at F1 ≈ 440, F2 ≈ 600 Hz

  • Pengzhou /o/ sound, which the graph says is at F1 ≈ 550, F2 ≈ 875 Hz

The corresponding artificial vowels are (unmute the audio in the imgur image)


Produced from auditoryneuroscience.com with a fundamental frequency of 110 Hz

  • 1
    Your answer is great with the images and the audios! As you can see in the graph, the vertical straps correspond to the amplitude frequency spectrum. The darker its color, the stronger its amplitude. The graph is acquired through Short Time Fourier Transformation. – Toosky Hierot Nov 28 '19 at 15:23
  • @TooskyHierot thank you, you're right that I should mention the darkness of the bands as corresponding to the peaks. This is not obvious to a non-engineering or science person. – dROOOze Nov 28 '19 at 15:35
2

具体说来,人的发音机制,大抵为肺部产生气压,由流体力学原理,声带周期性开合形成周期性声门波,经过声道调制形成输出声波。

这个过程可以视为脉冲幅度调制,即周期冲激信号经过成形滤波后再通过幅度调制系统。

对最终的声波行短时傅立叶变换,并将窗长取较大值(一般在两个基音周期以上)使频域波形混叠,能够得到每次变换时刻的声音频谱,将之作于横轴为时间,纵轴为频率,颜色深浅为幅度的图中,由于混叠假设,能够形成纵带状图样。

元音之分别,即在于其幅度谱包络之形状,以峰值表征,并循增序排列,将其共振峰标记为f1,f2(当然可能还有f3,f4等,但由人类听觉特别敏感的频率大约在几百至三千赫兹,所以频率更高的峰意义不如之前重大),以此为横纵坐标作图即得到你看到的图。

另外值得注意的是,除非有超强的绝对音感,想要听出共振峰的频率是不可能的;另外即便知道共振峰的分布,人(一般)不具有精确主观调动肌肉改变声带调制特性的能力——也即,以上的图并不能直接指导发音,顶多只是记录罢了。

The scientific terms are in bold font... If you are interested you may google them.

  • 1
    Your writing style isn't like the mainstream of modern Chinese. – dan Nov 27 '19 at 23:30
  • 術語用多的時候,可否提供些維基鏈接,提升答案的易讀性? – dROOOze Nov 28 '19 at 4:52
  • @droooze Sorry for that, because I learned it in class, here I just write down what I know...... – Toosky Hierot Nov 28 '19 at 15:04
  • It's hard to understand the whole graph without adequate math knowledge... In other words, I cannot decide put what links in my answer... – Toosky Hierot Nov 28 '19 at 15:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.