Modified 30 Mar 04 0012 hrs
| Answer | Correct Statement |
| 1. T | (Text)
Input is any data or instructions entered into the memory of a computer. (CIC) Input is any data received from an input device. The data may possible go directly to the CPU, or possible go directly to memory, depending on the computer system design. |
| 2. T | (Text)
An input device is any hardware component that allows users to enter data
or instructions into a computer. (CIC) An input device is a device that provides input. An input device is not necessarily operated by a user. For example, on a numerically controlled machine lathe in a machine shop, a laser used as a measuring device which takes data that is sent to a computer that generates control signals for cutting operation. This laser is an input device. |
| 3. T | The goal of ergonomics is to incorporate safety, efficiency, and comfort into the design of items in the workplace. [Note change in priority. During World War II, this discipline was called "human engineering", and was an important consideration in designing aircraft instrument panels, sonar consoles, and other devices.] |
| 4. T | A trackball is a stationary pointing device with a ball on its top. |
| 5. F (see explanation) | Audio input is the process of entering any sound into the computer, such as speech, music, and sound effects. Voice input is a type of audio input. However, "audio input" is a more general category. If you answered "true", you were partially correct. It just was not as general of a concept. |
| 6. F | A portable keyboard is a full-sized keyboard you can conveniently attach and remove from a PDA. |
| 7. T | Digital cameras typically allows users to review, and sometimes edit, images while they are in the camera. |
| 8. T | A video conference is a meeting between two or more geographically separated people who use a network or the Internet to transmit audio and video data. [The geographical separation could possibly be as close as adjacent rooms. The concept is that all communication occurs via the network. Video conferencing has been done in distance education via satellite simultaneously with multiple classrooms on different continents.] |
| 9. T | An optical reader is a device that uses a light source to read characters, marks, and codes, and then converts them into digital data that a computer can process. |
| 10. F | Optical Mark Recognition (OMR) is a technology that reads hand-drawn marks such as small circles or rectangles. [The concept is to use a low-resolution reader to sense the presence or absence of marks. The marks could have also been recorded by machine.] |
| 11. T | Biometrics is the technology of authenticating a person's identity by verifying a personal characteristic. |
| 12. F | Users with motor limitations have a variety of pointing devices available. A partial list: small track balls, head-mounted pointer, hand pad, foot pedal, facial motion sensor, gesture recognition, pneumatic control, bio implants to sense pressure or electrical activity, and combinations. |
1a. How does voice recognition work?
Voice recognition applies principles of signal processing, statistics, and artificial intelligence to the task of accepting an audio input signal and producing a sequence of tokens representing the information content of that input. The audio input signal may be supplied from a microphone directly, or replayed from a storage device. The sequence of tokens may be used to produce text for use by a word processor, commands to a computer system, data in a speaker identification file, or other uses.
The first stage of processing is to capture the input signals. In a conference room application, this may involve use of an array of microphones to form a beam to capture the voice signal of the speaker. A receiving array is used to reject signals and noise coming from undesired directions. The signal is then amplified, digitized, and processed to further eliminate noise and interference, and to extract tokens. Tokens are assembled into meaningful units or code that symbolically represent the meaning content of the speech.
You may have noticed that a person from New York City and a person from Louisiana sound different and use different idioms. The more you know about a signal, the easier it is to extract intelligence from it. For this reason, tokenizing and encoding speech can be easier if samples of signals representing known tokens are collected and modeled in advance.
The application of the speech processing system will determine the requirements. A system intended for public use cannot reasonably expect a speaker to spend several hours training the system before using it. Performance can be improved, however, by providing regional versions. Many public systems are specialized in purpose, such as an airline reservation system or telephone automated directory assistance, and do not need to recognize an unlimited vocabulary. By prompting the customer to supply short pieces of information at a time, the computation burden of tokenizing is reduced.
A general voice to text system for dictation must recognize a large vocabulary. A dictation system used by a physician or lawyer requires a very large vocabulary. The problem is simplified if the number of users are few. That makes it practical to require the user to read a selected list of words to train the software for that particular person's voice and speech patterns.
Another problem is dealing with homonyms. Context-sensitive software can reduce errors in selecting the desired word from a set of words that sound the same. The context can be grammatical or subject matter.
Faster processors, large memory, and digital signal processor (DSP) co-processors with large data cache help make the speech processing task more practical.
Some buzz words associated with speech processing include: adaptive signal processing, Fourier transform, wavelet transform, Karhunen-Loeve transform, Bayesian neural network, discriminant analysis, principal components analysis, Laplacian distribution. Lucent Laboratories (Formerly Bell Labs) is the premier research and development organization for speech and signal processing. Universities and government labs also continue to make important contributions. Examples of consumer speech recognition software include: IBM Via Voice, Dragon Naturally Speaking, QPointer Voice. Popular applications software now often have a voice-enabled version, such as Microsoft Office XP, Math Talk/Scientific Notebook
1b. How is discrete speech different from continuous speech?
Tokenizing speech is not trivial. Deciding where to break one word, or common phrase, and begin another from continuous speech is not a simple task. One simplification is to require a speaker to speak each word separately. While not a natural way of speaking, this permits recognition of a larger vocabulary.
Discrete speech recognition software requires the speaker to separate each word with a short pause. Continuous speech recognition software does not require each word to be separated with a short pause.
Expect discrete speech recognition software to more accurately transcribe spoken words to text for a dictation application than a continuous speech program would.
2a. What is needed to participate in a video conference?
You need a video camera, microphone, and a loudspeaker attached to your computer. You need video conference software. You should have a quite, acoustically dead room. It is sometimes helpful, but not required, to have other auxiliary input and output devices, such as a white board and scanner. A 3-dimensional ultrasonic scanner is useful for transmitting images of 3-dimensional objects.
2b. What is a whiteboard?
A whiteboard is a common electronic virtual sketchpad on which all participants can trade written or graphical information. A commercial version found in classrooms is called "SmartBoard".
3a. What is Optical Character Recognition (OCR) software?
Optical Character Recognition (OCR) software is a program that converts a bitmap image containing text images to character data that can be processed using a word processor.
3b. What is an image processing system?
An image processing system performs one or more of the following: record, amplify, store, subtract noise, feature recognition and extraction, image recognition and enhancement, modeling, manipulation.
4a. What is Optical Character Recognition (OCR), Optical Mark Recognition (OMR), and Magnetic Ink Character Recognition (MICR)?
Optical character recognition (OCR) matches the pattern of dots in a bitmap image against predetermined modeled characters. If the fit is good enough, the OCR software declares a match and records the character code (ASCII, EBCDIC, UNICODE) for that part of the image. The output of OCR software is text that can be edited by a word processor.
Optical mark recognition (OMR) detects the presence or absence of a mark. It does not attempt to determine the shape of the mark. OMR is often used for standardized testing. ScanTron is an example company that makes OMR readers for standardized testing. The advantage of using OMR rather than online testing is that expensive equipment is not required for recording responses of each student. A large group can be tested in a regular classroom. The disadvantages include:
|
Adaptive testing is cumbersome and impractical for large numbers of questions. |
|
OMR scoring is not as accurate as recording keyboard strokes. OMR readers can be very sensitive to marks, and may still misread a well-erased response as being still a valid mark. One common scanner reports the first mark detected on a response row. Wise students will always check the scoring of an OMR response form and not assume it was perfectly graded. Check the sheet immediately when you get it, and tell your teacher immediately of any suspected errors. |
|
Special answer sheets are needed. Some students make transcription errors when recording responses onto an OMR sheet from an answer book. |
|
If scanned responses are not captured and stored, but merely scored, the instructor must separately transcribe scores. |
Magnetic Ink Character Recognition (MICR) is character recognition applied to sensing and processing an image printed using magnetic ink. MICR also refers to a font style used with MICR. MICR is used on checks by the banking industry to automate the check-clearing process.
4b. How is each technology used?
Optical Character Recognition (OCR) requires an optical scanner to capture a bit map image of the document to be converted. Software attempts to convert all text area image segments into character data. Common OCR software partitions the image into graphics and text areas before conversion, and places text and images into a word processor document in approximately the same relative location. A spell checker identifies words not listed in its spelling dictionary for possible correction. In 1999, it took about 5 minutes to scan a single sheet using a flat bed scanner and correct spelling errors. It was common to have 5 errors (or more) per page.
A good typist could retype a single page of a document faster than scanning and doing corrections. The advantage of using OCR is apparent when the user is not a good typist, or if more than just a few pages are needed.
Optical Mark Recognition (OMR) requires an OMR scanner to capture the presence or absence of marks in predetermined locations on a form. The marks may be printed on the form or marked with a pencil. This is a very low resolution scanning process. The sheet is fed through a scanner. One type of scanner sends data to a file. Another type of scanner is first programmed by a form marked as a master. Subsequent forms are scored by the scanner. Results of the scoring is printed on the scored sheet by the scanner as the sheet exits.
Magnetic Ink Character Recognition (MICR) requires an MICR scanner to capture the image of a document imprinted with magnetic ink. The scanner does the conversion and passes the character codes to the processor. The accuracy of the conversion is high. MICR is used almost exclusively by the banking industry. Account and amount data are printed on checks in a predetermined position in a single line. The scanning process is simple compared to the OCR process. With similar constraints, modern OCR can perform with the same reliability.
5a. What is the Americans with Disabilities Act of 1990 (ADA)?
United States Code Title 42, Chapter 126, Section 12111 http://www4.law.cornell.edu/uscode/42/12111.html reports the provisions discussed on text page 5.35.
This act requires any company engaged in industry affecting commerce with 15 or more employees for each working day in 20 or more calendar weeks in the current or preceding calendar year to make reasonable attempts to accommodate the needs of disabled workers. Key provisions include:
"No covered entity shall discriminate against a qualified individual with a disability because of the disability of such individual in regard to job application procedures, the hiring, advancement, or discharge of employees, employee compensation, job training, and other terms, conditions, and privileges of employment."
"It may be a defense to a charge of discrimination under this chapter that an alleged application of qualification standards, tests, or selection criteria that screen out or tend to screen out or otherwise deny a job or benefit to an individual with a disability has been shown to be job-related and consistent with business necessity, and such performance cannot be accomplished by reasonable accommodation, as required under this subchapter."
"The term ''qualification standards'' may include a requirement that an individual shall not pose a direct threat to the health or safety of other individuals in the workplace."
"This subchapter shall not prohibit a religious corporation, association, educational institution, or society from giving preference in employment to individuals of a particular religion to perform work connected with the carrying on by such corporation, association, educational institution, or society of its activities."
"For purposes of this subchapter, the term ''qualified individual with a disability'' shall not include any employee or applicant who is currently engaging in the illegal use of drugs, when the covered entity acts on the basis of such use."
5b. How might gesture recognition and computerized implant devices help physically challenged users in the future?
Gesture recognition may help a person communicate via use of body motions. Computerized implant devices may help a person communicate by thinking. The communication can include control signals as well as thoughts.
1.a. Why is resolution important when using a scanner?
Answer: The resolution describes the sharpness and clearness of the digitized image. The higher the resolution, the better the image quality.
The scanned image is a bitmap image. The density of those dots is the resolution.
The resolution you scan at should be at least as high as the resolution you intend to print or process at. You can never regain lost details. The most you can do is to interpolate between adjacent sets of pixels. If you will be converting text image to text ASCII, you need a resolution at least as high as that required by the Optical Character Resolution (OCR) software.
Images can take up a huge amount of storage space. Always scanning at the highest possible resolution will result in rapidly using up your hard drive space. You will rarely need the highest possible resolution.
If you have the option of storing a document as text or as a bitmap image, choose text. A nearly 1000 page math book without graphics will fit on a single 3.5-inch 1.44 MB floppy disk. The simple map to an airport in the Windows 95 Lab manual drawn with MS Paint requires 1.4 MB. It is said that "a picture is worth 1000 words", but it takes up as much space as 200,000 words (estimated at 200 words per page). Instead of using pictures, we should make every 1000 words worth one picture?
1.b. How is resolution typically measured and stated?
Answer: Resolution is measured in dots per inch (dpi) by number of rows per inch (vertical dots per inch) and number of columns per inch (horizontal dots per inch).
3.a. What is a bar code?
Answer: A bar code is an identification code that consists of a set of parallel lines and spaces of different widths.
3.b. How are bar codes read?
Answer: The bar code is scanned by an optical scanner that detects the light patterns from the bar code lines.
3.c. On what products are bar codes used?
Answer: Bar codes are used on library books, blood units, mail, on articles in a warehouse for inventory management, retail food items in a grocery store, shipping containers used in the marine shipping industry, rail road cars, etc. In the grocery store, laser scanning of UPC codes became common around 1982. This resulted in increased accuracy of billing and speed of checkout at the cash register, and much improved inventory control.
3.d. What is the difference between a bar code and a Web bar code?
Answer: A Web bar code is a bar code used by advertisers to identify a product and fetch a corresponding Web page. The CAT optical scanner given away at Radio Shack is used to read such a code, translate the code to a URL, and therefore fetch the associated page.
The picture in the text shows a Web bar code that is slanted. The CAT optical scanner does not require that the bar code be slanted. The scanner only detects changes in light intensity as the scanner passes over the code. It does not know that the code is slanted. It only cares that the lines are parallel, and that the user scan across the code at approximately a uniform speed.
4.a. How is speaker-dependent software different from speaker-independent software?
Answer: Speaker-dependent software uses a pre-recorded profile of your voice to decode speech you enter at a later time. Speaker-independent software matches your input against a pre-determined library of word patterns.
Speech recognition is a mathematically challenging problem. Expect speaker-dependent software to perform better as input to dictation software than speaker-independent software. Think of regional differences in speech. You may have noticed that a person from North Carolina sounds different than a person from Ohio. If you limit the vocabulary to a prescribed set of words, it is more likely you can model how to detect these spoken words robustly. Speaker-independent software is useful for airline reservation systems, telephone directory information service, and other applications serving a national population.
5.a. What is videoconferencing?
Answer: A videoconference is a meeting between two or more people in separate locations by transmitting audio and video data to receivers at each site.
The communication links can include internet or satellite transmission. It can be between inmates and visitors within the same building. The communications links are the expensive part of videoconferencing. The cost is justified when compared to travel costs and associated interruption of other work.