4. the speaker makes correct utterances. The

 

4. SPEECH ACQUISITION

First step in this
level is to select good speakers. A notice was published among the departments
of Islamic University inviting speakers in this regard. Huge amount of
interested students both male and female were applied to do the job. An
audition was arranged to check their comparative efficiency of correct
utterance/pronunciation. Depending on their performance, 75 male and 75 female
speakers were selected to finalize the speaker list. The major content of the
list was the students of language and computer related departments. The
selected speakers were very young with age range of 18-25 years. Finally
selected speakers were attended a two day workshop. The objective of the
workshop was to concern the speakers about the theory, methods and work plan of
the project. The speakers were given practical training of speech acquisition
such as headset setting, loudness of utterance etc.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!


order now

Speech data were
recorded in Laboratory environment by close-talking microphone directly
connected to the computer. The speakers were asked to read the text in standard
pronunciation as well as possible. The whole recording was done in the presence
of the researcher who controlled the recording. The control includes running the
recording script, starting the recording session, breaking the recording,
playing back the recorded speech, re-recording the sentence if required and
moving to the next sentence. Every speaker was explained the purpose of the
project and instructed to start the recording when she/he was ready to read.
After the entire session for a reader was finished, all the utterances were
listened to by the reader and the researcher for corrections. Furthermore, as
only 8-10 speakers have been scheduled to read per day, the researcher was able
to listen again to the recorded speech at the end of each day to control its
quality. There were some console operators to help the researcher and when an
operator identifies the hesitation or erroneous utterances, he asked the speaker
to repeat the utterance until the speaker makes correct utterances. The
recorded speech data were taken with 8 kHz sampling with 8 bit quantization.
The recorded speech was stored as wav file format in various lengths
depending on the speaker’s ability to speak over a length of continuous time. Connected Words were uttered with a
minimum pause between two consecutive words. But the speakers were very much
conscious about the context dependence of word utterances. Thus the final
result was that the speaker speaks normally but with a word-short pause-word
style which ensure a minimum separation between two consecutive words.
Depending on the speaker’s ability, the total text was recorded and stored as wav
file format in two or three primary unedited files. The speakers were trained
and directed continuously to maintain the similar rhythm in text reading to
ensure the same utterance of text.

5. SPEECH EDITING AND LABELLING

It is necessary to
check the recorded data from the following points: difference between the utterance and the utterance list, degree
of dialectal accent, speech rate, clarity of pronunciation, recording level,
noise, etc. Especially, we found some speakers speak with very low speech level
and it was possible to magnify the amplitude of such speech data to a suitable
level. However, the recording was carried out in an environment, which was not
truly noiseless. The recoding instruments were also produced a little noise in
some cases. All types of problems were identified and corrected during editing
phase. Noiseless clean speech files were separated from the noisy speech files.
Noisy files were cleaned using various filters and tagged with a comment. The
original unedited recorded files were edited so that each file contains one
complete sentence. As the HMM Toolkit developed by Cambridge University
Engineering Department was already proved its efficiency and using frequently
by most of the research workers, the database was labeled by following the
specified format of speech data to make it ready for use in HMM Toolkit for
evaluation 11. Two edited and finalized files are shown in figure-1. As shown
in the figure minimum length file contains 3 words with duration of 2.3 seconds
and maximum length file contains 11 words with duration of 8.5 seconds. Thus
the estimated average time required for each file was 5.4 seconds. The total
recording time for connected words database was about 62 hours after editing
and labeling.

x

Hi!
I'm Joan!

Would you like to get a custom essay? How about receiving a customized one?

Check it out