i. Discuss how the Digital signal processors are different from other general purpose microprocessors in terms of Architecture, Applications, Specifications (Memory, speed, I/O, Buses, Power requirements, Instruction Set, software development
tools … etc).

ii. Compare (using a table) between the specifications of one of the DSP processors and one of the general purpose processors from the same generation, such as TMS320C6201 and Pentium MMX

Answers:

Introduction

Signal processing using digital computers and special purpose digital hardware has taken on a huge significance in the recent years. There are hardware engineers who use the term digital signal processor to refer to DSP while the algorithm developers refer to it as the digital signal processing. The literature review section analyses the digital signal processor in terms of its architecture, features, and algorithm as compared to the general-purpose microprocessors. Digital signal processing is applicable in sound and harmonics, communication devices, automotive industry, medical applications, military applications, image and video applications as well as in process control in the field of mechanics (Monica, & et al, 2011).

Literature Review

The inception of digital signal processing has its background based on the limitations of analog signal processing. The analog signal processing had the following limitations: -

• Sensitivity to electrical noise
• Limited dynamic range for voltage and currents
• Inflexibility to changes
• Difficulty while implementing non-linear operations and time-varying operations
• Difficulty in information storage (Vijayalakshmi, & et al, 2011)
• Limited repeatability due to tolerances and changes in environmental conditions
• Limited accuracy due to component tolerances and undesired non-linearities

General purpose microprocessors

These microprocessors perform the basic functions of the standard computer’s CPU. They run on a few integrated circuits and operate the ALU and control Unit. They are not designed to perform specific tasks unlike the digital signal processors. They also operate digital signals and have ADC at the input and DAC at the output. The converters are necessary in that they allow analog signals to be input, the signals are converted to digital form for processing, and back to analog form for output on other peripheral devices. The microprocessor is not dependent on a given language or software and they are commonly found in the personal computers. When compared the DSP processors, the general-purpose microprocessors tend to be slower in operation (Mukesh, & et al, 2009).

Digital signal processors

The DSP processor is a specialized type of microprocessor. Digital signal processing involves the capture of a detector signal using a continuous running digitizer. The information contained in the pulse shapes is used to trigger the data acquisition system. A lot of information is contained in the signal shape and the pulse shape information can be written to a data file for off-line analysis using a variety of tools. The digital signal processor preserves the information provided by the equipment used. It simplifies the hardware requirements for the advanced lab since changes in signal processing only requires change in the analysis tools. The processors adopt the approach of digitization first, then the signal processing online, off-line, or both. The information obtained is preserved as deemed fit by a user. The signal information that is captured involves the entire waveform, pulse height, and the time of arrival. 

The aim of the design of the DSP processors is to perform high-performance, complex computations which are repetitive and numerically intensive. DSP processors incorporate: -

• RISC-like architecture
• Harvard-Bus architecture
• Regular instruction cycle
• Parallel processing
• An internal memory organization
• Multiprocessing organization
• Local links
• Memory banks interconnection

The DSP processors often fits a single instruction, multiple data framework also known as SIMD. The Intel, AMD, and Motorola have DSP-like instruction extensions on their processors such as the Pentium MMX. Unfortunately, most of these do not support a true MAC with guard bits.

Table 1 Comparison of GPP and DSP microprocessors

System Functionality

The real time digital audio processor system functionality:

• 5-band graphic equalizer for high quality stereo music signal (48kHz 16-bit A/D)
• Elimination of two interference tones present at 5kHz and 10kHz
• 5-band equalizer controlled by a GUI running on the PC.
• Use minimum DSP resources (processing time and memory)

Discussion 

Digital signal processing allows for repeatability and is flexible hence can be achieved with software implementations. The digital storage is cheap and the digital information can be encrypted for security. Unfortunately, the sampling of the input signals may cause a loss of information as well as a limited speed of the processors. The conversion from analog to digital and back may require the use of mixed-signal hardware. The digital audio system converts a continuously changing waveform into a series of discrete levels which is the digital form of the signal. The signal can be mathematically manipulated in its digital form as opposed to the analog form affect noise and other disturbances are filtered (Wolfs, et al., 2005).

The TI TMS320 is a member of the Texas Instruments TMS320 family innovated in the 20^th century. Its architecture contains 16-bit word size, single 32-bit accumulator with D & P data buses, 5 aux registers for data or address with no modulo support. It has an integrated MAC/ ALU with no parallel computation. 

The TMS320C62XX series has separate internal program and data spaces. The program operates as,

Table 2 Program Operation Information

While the data,

Table 3 Program Data

The Intel Pentium MMX, supports the MMX technology. It is compatible with large software Base. 32-bit processor with 64-bit data bus. It uses a superscalar architecture with enhanced pipelines and separate code and data caches. The MESI Cache Protocol and the 16-Kbyte code with write back data. It has multi-processor support with the IEEE 1149.1 boundary scan.

Conclusions

In a nutshell, the real time digital audio system is an implementation on LabVIEW that demonstrates the application of advanced DSP in I/O at the audio system design and the complex manipulation of signals and elimination of noise. 

References 

Wolfs, F., Alexander, M., Miner, D. & Skulski, W., 2005. Using Digital Signal Processing in the Advanced Laboratory, University of Rochester, Rochester, NY 14627: Laboratory for Laser Energetics.

Anushiya Rachel P. Vijayalakshmi and T. Nagarajan "Improving Speech Intelligibility in Cochlear Implants using Vocoder-Centric Acoustic Models" in IEEE/ICRTIT 2012 pp. 66-71.

K.I. Kirk L.S. Eisenberg A.S. Martinez and M. Hay-McCutchen-"The Lexical Neighborhood Test:-Test-Retest Reliability and Inter-list Equivalency"-Progress report no.22 (Indiana university).

J. Markel "The SIFT Algorithm for Fundamental Frequency Estimation" IEEE transactions on Audio and Electroacoustics Vol. 20 Issue 5 Dec. 1972 pp. 367-377.

Douglas Reynolds "Gaussian Mixture Models" MIT Lincoln Laboratory.

Rosenberg A.E. 'Effect of Glottal Pulse Shape on the Quality of Natural Vowels' The Journal of The Acoustic Society of America 1970 pp. 583-590.

R. Rabiner and R.W.Schafer 'Digital Processing of Speech Signals' Pearson Education 2004.

T. Nagarajan T. Monica "Segmentation of Speech Signal into Phonemes using Two-Level GMM Tokenization" IEEE ICRTIT 2011 pp. 843-847.

Ra. V. Jayanthan P. Vijayalakshmi and P. Mukesh Kumar 'Auditory model based acoustic CI simulations for patients with profound hearing loss' International Conference on Implantable Auditory Prosthesis (CIAP) 2009 pp. 228.

Vijayalakshmi P Nagarajan T and Preethi M 'Improving speech intelligibility in cochlear implants using acoustic models' WSEAS Transactions on Signal Processing Issue 4 vol. 7 October 2011 pp. 103-116.