البصمة الصوتية : أمد بداية التصويت أنموذجا

Abstract EN

Utilizing the right acoustic cues in speaker identification cases is one of the challenges that face speech analysts.

A speech analyst usually uses spectrograms to search for acoustic cues in the speaker identification process.

A spectrogram shows the acoustic signal in a three dimensional graph (frequency, intensity and time).

An acoustic cue can be related to the frequency, intensity or time.

A combination of different acoustic cues would help a speech analyst to come up with the right decision.

However, the frequency and intensity of a speech acoustic signal can be seriously affected by background noise and/or reverberation.

A mask, for example, would dislocate the format frequencies of vowels (Figure1).

Figure 1.

The long low vowel / a / produced by the same speaker in a quiet environment (right) and while wearing a mask (left).

SHAPE \ * MERGEFORMAT Temporal acoustic cues are useful when analyzing speech especially when its spectrum is somehow affected.

One of the most salient temporal acoustic cues is voice onset time (VOT).

It is the time between the release of articulators and the onset of the vocal fold vibration.

For example, in the case of / p / production in a word such aspeak, VOT is the period between the release of the two lips (end of closure) to the period that the vocal folds start vibrating to produce the vowel.

Since VOT is a result of a complicated timing procedure where the glottis coordinates with supraglottal organs, it has been found that speakers do not consciously control the VOT duration.

VOT has attracted researchers from different backgrounds: language typology, language acquisition, psychology, phonetics, second language effect on the first language and speaker identification.

The results of these studies and other similar studies contribute to the human knowledge in their related fields.

The objective of this paper is to investigate the role of VOT in speaker identification and the effect of the second language on the mother tongue language.

The subjects consist of 16 Arabic speakers who are native Saudis.

8 of the speakers (Group I) have less contact with English as a second language, and the other 8 speakers (Group II) have studied, lived and obtained their Ph.

D.

degrees in an English speaking country.

The target consonants are the alveolar voiceless stop / Ø, the velar voice- less stop /„Ø and the emphatic alveolar†voiceless stop / ◊ Ø Æ The first two consonants are in both the Arabic and English consonant inventories.

The last consonant occurs only in the Arabic consonant inventory.

The consonants were written in the carrier CV-VC where C is / z /, V is / a / and - is the target consonant.

A list of the consonants within their carrier token was made where each consonant is repeated 6 times and the list is randomly made.

Not two tokens of the same consonant were in sequence.

The results of the recording are: 3 consonants X 6 repetitions X 16 speakers = 288 tokens.

The recording was made using the Kay Elemetrics Multi-Speech System (Model 3700) in a sound-proof chamber in the Speech Center at King Abdulaziz City for Science and Technology.

The same system was used to display the spectrograms of the tokens.

VOT and closure duration were manually measured.

The results of the measurements were statistically analyzed.

The results of the study show that speakers differ in terms of VOT production in the three consonants.

That is, speakers produce the target consonants with different VOT duration ranging in average from 21 milliseconds (ms) to 78 ms for / Ø, ≤± ms to 55 ms for / „Ø and 14 ms to 33 ms for Ø◊ØÆThe speakers also differ in terms of closure duration ranging from 51 ms to 81 ms for / Ø, 59 ms to 84 ms for /„Ø and 62 ms to 93 ms for Ø ◊ Ø Æ The two groups show different measurement results.

Group I show VOT longer in average than that of Group II when the shared consonants ( / © Ø„ are produced.

This means that Group II which is more exposed to English show shorter VOT when producing the sounds that are present in the two languages, Arabic and English.

Hence, English has a negative effect on Arabic VOT of the shared consonants.

This probably is due to the fact that native English speakers show longer VOT than native Arabic speakers when producing voiceless stops in their languages.

Therefore, Arabic native speakers who learn English as a second language show, unconsciously, that the two languages are pho- netically different by maximizing their VOT difference.

The result is a shorter VOT for Arabic voiceless stops that are shared in the two languages, but the VOT of the non-shared stop is unaffected.

Many of the studies on bilingualism focus on the effects of the first language on the second language, what is called language transfer.

For example, the phonological transfer of the Arabic / b / by Arabic native speakers when they speak English words with the / p / consonant.

How-ever, only a few studies have been conducted on the reverse effect, i.

e.

the effect of the second language on the first language.

In summary, the results of this study show the presence of individual differences among Arabic speakers in terms of VOT variations.

The results also show that a phonetic difference between the first language and the second language is maximized when the speakers are more fluent in the second language.

Thus, it is found that a fluent Arabic speaker of English produces Arabic voiceless stops with shorter VOT than a non-fluent Arabic speaker of English.

This means that it can be predicted from Arabic speech that the speaker is fluent in a foreign language with long VOT such as English.