**6. Concluding remarks**

12 Computational and Numerical Simulations

0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02

<sup>7</sup> x 10−3

0.5

1

1.5

Amplitude

2

2.5 x 10−3

Amplitude

Amplitude

F0 = 215.4 Hz

430.8 Hz 646.2 Hz

<sup>0</sup> <sup>200</sup> <sup>400</sup> <sup>600</sup> <sup>800</sup> <sup>1000</sup> <sup>1200</sup> <sup>1400</sup> <sup>1600</sup> <sup>1800</sup> <sup>2000</sup> <sup>0</sup>

1068 Hz

F2 = 1497 Hz

F2 = 1486 Hz

<sup>0</sup> <sup>500</sup> <sup>1000</sup> <sup>1500</sup> <sup>2000</sup> <sup>0</sup>

F1 = 969.2 Hz

Temporal segment: 0.3715 s − 0.4644 s

Frequency (Hz)

Temporal segment: 0.835s − 0.928 s

<sup>0</sup> <sup>200</sup> <sup>400</sup> <sup>600</sup> <sup>800</sup> <sup>1000</sup> <sup>1200</sup> <sup>1400</sup> <sup>1600</sup> <sup>1800</sup> <sup>2000</sup> <sup>0</sup>

F2 = 1496 Hz

F2 = 1495 Hz

F1 = 948 Hz

Frequency (Hz)

Temporal segment: 1.0217 s − 1.115 s

<sup>0</sup> <sup>200</sup> <sup>400</sup> <sup>600</sup> <sup>800</sup> <sup>1000</sup> <sup>1200</sup> <sup>1400</sup> <sup>1600</sup> <sup>1800</sup> <sup>2000</sup> <sup>0</sup>

F1 = 934 Hz

Frequency (Hz)

(f)

(d)

(b)

0.005

0.5

1

**Figure 7.** Spectra of different temporal segments after applying the sliding window procedure of a well-pronounced vocal number 5 = / :/ by a woman of 29 years old.. (a) 0 - 92.8 ms, (b) 0.371 - 0.464 s, (c) 0.464 - 0.557 s, (d) 0.835 - 0.928 s, (e)

2

Amplitude

3

x 10−4

1

1.5

Amplitude

2

2.5

<sup>3</sup> x 10−3

0.01

0.015

Amplitude

0.02

0.025

Temporal segment: 0 − 92.8 ms

851 Hz

Frequency (Hz)

(a)

Temporal segment: 0.4644 s − 0.5573 s F1 = 936 Hz

<sup>0</sup> <sup>200</sup> <sup>400</sup> <sup>600</sup> <sup>800</sup> <sup>1000</sup> <sup>1200</sup> <sup>1400</sup> <sup>1600</sup> <sup>1800</sup> <sup>2000</sup> <sup>0</sup>

Frequency (Hz)

Temporal segment: 0.928 s − 1.021 s

<sup>0</sup> <sup>500</sup> <sup>1000</sup> <sup>1500</sup> <sup>2000</sup> <sup>0</sup>

F1 = 947.7 Hz

F2 = 1486 Hz

Frequency (Hz)

(e)

0.928 - 1.021 s, (f) 1.021 - 1.115 s

(c)

In this paper we have improved the tool implemented in [1], which consists in a software system for the teaching of English phonology. Pavón's contribution allows phoneme recordings, which are later on compared to similar sounds in the system. However, it offers a comparison based on the time domain, which is certainly not significant when providing help for learning a second language pronunciation. Moreover, it includes female voice recordings only, so male users (and children) would not obtain a significant result. Taking into account that Pavón's original idea is very good for those students who lack listening and pronunciation skills, this paper describes a new procedure to be added to the previous system and which is based on a frequency domain analysis. In this way, by means of a formant detection algorithm based on [20] and [11], the system can offer a more realistic contribution to the teaching of English pronunciation and phonology. F1 and F2 indicate oral cavity opening and tongue position respectively, and so the system specifies whether students have to open or close their mouths and which tongue part must be particularly employed in each vowel sound. As [1] makes use of female voice recordings only, our subjects are female adults. However, our formant detection algorithm would work with male and children voices equally. Male and children native' speakers are required for reference in order to have their voices recorded and can be employed to appropriately compare with male and children non-native users of our system.

10.5772/57221

343

http://dx.doi.org/10.5772/57221

[10] Awrejcewicz, J. Numerical investigations of the constant and periodic motions of the human vocal cords including stability and bifurcation phenomena. Journal of

Spectral Study with Automatic Formant Extraction to Improve Non-native Pronunciation of English Vowels

[11] Barbancho-Pérez, I., Jurado-Navas, A., Barbancho-Pérez, A., Tardón, L. Transcription

[12] Jurado-Navas, A., Puerta-Notario, A. Generation of correlated scintillations on atmospheric optical communications. Journal of Optical Communications and

[13] Jurado-Navas, A., Garrido-Balsells, J. M., Castillo-Vázquez, M., Puerta-Notario, A. A computationally efficient numerical simulation for generating atmospheric optical scintillations. In: Awrejcewicz J. (ed.) Numerical simulations of physical and

[14] Jurado-Navas, A., Garrido-Balsells, J. M., Paris, J. F., Castillo-Vázquez, M., Puerta-Notario, A. Impact of pointing errors on the performance of generalized

[15] Taillefer de Haya, L., Silva Ros, M. T. New technologies in English Applied Linguistics. In: Proceedings of the 30th International AEDEAN Conference, Huelva, Ed. María

[17] Chiba, T., Kajiyama, M. The Vowel: Its Nature and Structure. Tokyo: Kaiseikan; 1941.

[18] IPA, The International Phonetic Association. http:www.langsci.ucl.ac.ukipa (accessed

[19] Brown, J. C. Calculation of a constant Q spectral transform. Journal of the Acoustical

[20] McCandless, S.S. An algorithm for automatic formant extraction using linear prediction spectra. IEEE Transactions on Acoustics, Speech, and Signal Processing

[21] Oppenheim, A.V. Discrete-Time Signal Processing. Upper Saddle River, New Jersey,

atmospheric optical channels. Optics Express 2012; 20(11) 12550–12562.

[16] Fant, G. Acoustic Theory of Speech Production. The Hague: Mouton; 1960.

of piano recordings. Elsevier Applied Acoustics 2004; 65(12) 1261–1287.

Dynamics and Stability of Systems 1990; 5(1) 11–28.

engineering processes Rijeka: Intech; 2011. p. 157 - 180.

Losada Friend et al. Huelva: U de Huelva, 2007.

Society of America 1991; 89(1) 425–434.

USA: Prentice-Hall, 2nd edition; 1999.

Networking 2009; 1(5) 452–462.

20 May 2013).

1974; 2(2) 135–141.
