Questions and answers about SiCECA
*Shiomi theory referred here is the theory developed by Dr. k. Shiomi of ENRI, Electronic Navigation Research Institute Japan, entitled New uttered voice analysis using chaos theory to evaluate degree of activity of human brain.
SiCECA in general
Q1: What does SiCECA stands for?Technical questions
Q1: What is the input data to SiCECA?Q1: What does SiCECA stands for?
A: It stands for Shiomi Cerebral Exponent
Calculation Algorithm.
Q2: What is SiCECA in general?
A:
SiCECA is a high-speed signal processing computing algorithm specialized for
human uttered voice to evaluate chaotic nature of the voice and its chaotic
randomness. The software implementing this algorithm is able to compute chaos
and chaotic randomness from a human speak. We believe that the firmness of
uttering functions of the speaker brain is evaluated from the level of chaos
and brain activity level is evaluated from the level of chaotic
randomness.
Q3: What is the Shiomi’s theory?
A:
Shiomi theory is the theory that SICECA (computing algorithm) is based on.
Originally the Shiomi theory was developed for aiming to measure the degree of
tiredness of pilots of aircraft and air traffic controllers without disturbing
their work. Dr. Shiomi found that there were strong relation between human
tiredness and chaotic randomness of his/her voice. Dr. Shiomi named brain
activity level to the chaotic randomness of human voice, and developed a
high-speed algorithm SiCECA to compute the value of brain activity
level. Comparing with other schemes to measure tiredness of human being, the
advantage of the Shiomi theory is that it does not need to disturb human
working to measure his/her tiredness because it uses human voice while working.
In contrast, most of the other schemes require interruption of human work to
measure his/her tiredness.
Q4: Could you provide me in more detail about the Shiomi’s theory?
A:
Dr. Shiomi found that there were differences in chaotic nature of uttered voices
between humans in calm condition and under nervous or tensioned condition. Based
on the Shiomi theory, by numerically representing the degree of chaos, it is
possible to measure the degree of activity of the human brain. By continuously
monitoring the brain activity, accumulation of tiredness is possible to detect.
By analyzing an uttered voice in a chaotic method, an index value of noise which
disturb the chaos system of the uttered voice shows positive relationship with
the stress suffering to the voice owner. The index is named the value of
brain activity level or CEM (cerebral exponent macro), corresponds to the
degree of brain activity of the speaker and the value increases in stressed
conditions. Judging from his experimental results, it could be said that the
cerebral exponent macro (CEM) has relatively small differences among individuals
so much as the case of blood pressure and does not depend on languages.
Therefore CEM would be possible to use as a general measurement of reaction to
stressor given to human.
Q5: What is the human brain
model which describes the mechanism of relation between the brain activity level
and stressor? What is happened in human brain?
A: It is still unknown what
happens in human brain exactly, but in this moment, the following brain model is
thought as the reason why the values of brain activity level changes in stressed
condition.
It is known that the Wernicke area and the Broca area of a brain play
major roles in human speak. The Wernicke area is at the left side of lobe and
functions to understand language. The Broca area in the frontal lobe controls
the muscle of uttering organs. When a human thinks and wishes to speak, a
sentence is generated in the Wernicke area and transmitted to the Broca area
through angular gyrus and gyrus supramarginalls. The Broca area then receives
the signals from the Wernicke area, merges it with emotion from the right brain,
and controls the muscle of uttering organs. As a result, a human being generates
voice that is meaningful sound of a human being. In the Shiomi model, it is
thought that a lot of information signal would be simultaneously exchanged among
various parts of a brain, and there would be signal interferences inside,
outside and between the Wernicke area and the Broca area. Such interferences
result in chaotic noise included in uttered voice. When we are relaxed and
receiving less stress or we are absentminded. The amount of information
exchanged among brain parts would be relatively less and uttered voice includes
less amount of noise. In stressed condition, the information exchanges
increases, and the noise also increases. Note that the noise mentioned above is
not the noise we are used to know and we can hear in voices, but the noise that
disturbs chaotic nature of the voice. SiCECA quantitatively computes the level
of noise that disturbs the chaos system of the uttering voice.
Q6: What are the objectives of SiCECA?
A:
The objective of SiCECA is to realize the equipment and method to quantitatively
evaluates the human brain activity level in real time.
Q7: How does SiCECA measure
tiredness or fatigue from speak?
A: SiCECA measures the brain
activity level. Tiredness and fatigue can't be measured directly by SiCECA.
However, it is observed that when a person is tired, by giving the person short
time of brain workload, the value of brain activity level (CEM value) is quickly
increased, so by continuous monitoring, increase of the value of brain activity
level is observed while a person is getting tired.
Q8: Is it possible to measure
the intelligence of a person by SiCECA?
A: No,
the value of brain activity level is not the measurement of clever or poor in
intelligence.
Q9: Is it possible to measure how much a person is tired?
A:
Measurement of degree of tiredness is one of our major subjects since our early
days of research. However, we still need more work to achieve a simple and
accurate measurement of degree of tiredness. In this moment, from our
experiment, we observed increase of the value of brain activity level by a short
time brain-workload in the case a person tiredness is being accumulated. For
an example, 30 seconds of reading of a book by a tired person shows a high value
such as over 500 of brain activity level. Even though a high value of brain
activity level does not necessarily mean that the person is tired, it could be
said that if such the high values of brain activity level continuously keep on a
long time such as 10 to 30 minutes, most likely the person feels tiredness or
the person will eventually (in a half an hour) feel tiredness. There is a
difficulty that the personal feeling about tiredness is quite depended on
individuals. Some people may judge themselves as not tired even though his/her
tiredness is obvious. So declarations of individuals about his/her own tiredness
are not always correlative with the value of brain activity level. The brain
activity level is an objective criterion and not controllable by the person. In
contrast, personal declaration is a subjective criterion.
Q10: How the value of brain
activity level from SiCECA is changed when tired or relaxed?
A: In
general, the value of brain activity level is around 500 in average for male.
Females show relatively lower values. In case the brain is actively working, the
value is getting higher value. When a person is getting tired or reading/talking
about difficult subjects, the value exceeds over 550. When a person is relaxed,
the value is low around 350 to 450. Judging from our experimental results, when
a person is getting too much tired, the value also goes down to lower value such
as 350 to 450. We call this burn out.
Q11: Is it possible to measure the trembling of human mind or tension from
human voice?
A: Yes, when we analyze uttered voices in continuous
manner, fluctuation of the value of brain activity level is observed. Our
understanding is that this would be the result of fluctuation mind or mental
strain. However, the fluctuation of the value is so small as the environmental
noise and setting of a microphone affects a lot on the value.
Q12: How about language
dependency of SICECA? Is SiCECA applicable only for a specific uttering
language, or applicable for any language?
A: SiCECA algorithm does
not depend on languages, and applicable to any uttering language.
Q13: Is it possible to
measure tiredness of animals from their voice?
A: No, SiCECA is for
human voice and based on a human brain model. Because animals do not have highly
developed language, SiCECA would not work on animal brain activity.
Q14: Has a patent being given to the algorithm of SiCECA?
A: Patents for
SiCECA and a few applications of SiCECA are being filed in international.
Q15: How can I get more detail information about SiCECA?
A:
Refer to technical information and experimental result at
this web site.
Q16: What is the future
direction of SiCECA?
A: One area we are focusing our
effort is to find relationship between resolution and reliability. Currently,
the value of brain activity level is swayed by recording equipment and
conditions. We would like to learn about resolution of the value of brain
activity level in each recording cases, and a way to improve reliability of the
value. For an example, we would like to identify in what cases the difference of
420 and 430 of brain activity level is meaningful, and in what cases, the
difference is meaningless. Also, we would like to improve the computing
performance of SiCECA, and to improve the algorithm to cope with environmental
noise. We also plan to develop application software products to apply SiCECA in
practical use as a human factor management system.
Q17: Can I try SiCECA?
A:
Yes, this web site provides you with Free SiCECA Services. You can
try SiCECA with your voice in free of charge.
Q18: When I have more voice
data than Free SiCECA can handle, what is the best way?
A: SiCECA premium service is the best suited to such the case.
SiCECA premium service is a high performance, limit-free, and
charged based SiCECA computing service through Internet.
Q19: Is there any commercial software product that implements SiCECA
algorithm?
A: Yes. Parole is the one. Parole is Windows
PC based, SiCECA software product available now. For more information, please
contact us at info@siceca.org .
Q20: How far the extent of
corroborative test being made for SiCECA?
A: We have conducted
sampling and analyzing of 5 seconds to a minute long of voices from hundreds of
people. We also have tested voice sampling over television broadcasting. We are
now conducting corroborative tests with many research organizations and
companies in various areas.
Q21: How can I conduct a
collaborative study with the SiCECA research organization?
A: We are
welcome to have a contact from you at info@siceca.org.
Q1: What is the input data to SiCECA?
A:
An input data for SiCECA is a WAV format file of PCM, 48KHz, 16 bits, monaural.
Q2: How to produce a voice file for SiCECA?
A:
You can use any voice recording software available in free or commercial based
which can record voice in a WAV file.
Q3: Is there any limitation on the length (in seconds) of voice files?
A:
No, but a voice files of longer than 2 seconds and shorter than 10 seconds
might give you a proper SiCECA result.
Q4: Can I use a voice data recorded from television? How about the voice over
the telephone or MP3 and compressed voice data?
A: Yes, you can use
voices from television. However, compressed voices such as voice over a
telephone or cellular is not suitable for SiCECA. The availability of compressed
data such as MP3 is the one of our challenges in this moment.
Q5: Can I use a low cost microphone to record voices for SiCECA, or should I
use a special, expensive microphone? Can I use a built-in microphone of PC?
A:
If your purpose of using SiCECA is just to see your fatigue from 5 to 10 seconds
of your uttering voice, a low cost microphone attached or connected to your PC
would be good enough. In such the case, you should care not to pick up fan noise
generated by your PC. However, if your purpose of using SiCECA is more serious
and you want accuracy, you should be careful to choose microphones. We recommend
microphones which is able to cover frequency range from 80Hz. This is because
SiCECA is strongly affected by the voice signal pitch frequency bands. Even in
the case the frequency range noted on the product specification sheet of microphone is said to cover
from 150Hz, if the microphone does not have any low cut filter in it, you can
use the microphone for SiCECA. In the case you want to evaluate how the values
of brain activity level vary in time series, It is highly recommended to use the
same microphone for all of recordings to eliminate fluctuation of characteristic
of microphones.
Q6: In order to obtain meaningful
numerical values, what would be the minimum length of voice we have to use?
A:
Effective values could be obtained by 5 seconds of voices.
Q7: What factor determines the
processing time of SiCECA? Is it the size of input data?
A: Not only
the length of time of voice file but also the sampling rate of the voice file
determines the processing time. If the sampling rate of the voice is doubled,
the amount of processing is also doubled. Based on our rough estimates, it takes
approximately 10 times processing time of the uttered voice time (second) in the
case of 44.1KHz sampling rate (i.e., if the voice is a 5 seconds period, the
processing time would be about 50 seconds) by the current PC.
last update 07/03/05