Tutorials to .com

Tutorials to .com » Mechine » Digital » LabVIEW development platform based on the design of the sound energy

LabVIEW development platform based on the design of the sound energy

Print View , by: iSee ,Total views: 23 ,Word Count: 2152 ,Date: Wed, 26 Aug 2009 Time: 10:10 AM


Voice signals is that people communicate and exchange key media. Voice has the double property, one voice has the semantic function; the other hand, voice, after all, a voice, which is produced by the human mind through a set of ideas to control the vocal organs nerve signals into the air vibration signals , and then by air to pass to the person's ear or the receiver of the signal. The basic role of the voice exchange of information constitute a voice communications. In voice communications systems, voice signal transmission storage and processing methods are diverse. Processing of speech sounds in general can be divided into the following areas: speech analysis, speech enhancement, speech coding, voice synthesis and voice recognition and understanding. Sound equipment, a single from the record view, there gramophone, tape recorder to the present digital MP3 players, of which about recording technology has also been rapid changes. Thus, in a highly developed information society of today, with digital method of voice transmission, storage, identification, strengthening is particularly necessary.

LabVIEW development platform

LabVIEW is the only compiler-based graphical programming language, use the "WYSIWYG" and visualization technologies to build man-machine interface, with icons that function modules, using icons to represent the connection between the various modules data transfer between. At the same time, LabVIEW has inherited the structure of high-level programming language and the advantages of modular programming, support modular and hierarchical design, such structures are designed to enhance the readability of the program, its interface is very intuitive image.

Virtual instrument based on computer as the hardware device support, full use of the computer's computing, storage, retrieval, display and document management features such as the traditional instruments of specialized function software to make it integrate with the computer integration, thus constitute an appearance to Taiwan from the features are exactly the same as traditional instruments, while the full access to the computer resources of the instrument system. Traditional instruments usually signal acquisition, signal analysis, signal output of three parts; virtual instrument can also be divided into data collection, data analysis and processing, display the results of the three major functional blocks.

Virtual instrument system components

Virtual instrument system is composed of computers, hardware platforms and application software composed of. From the construction method, there are data-acquisition card (DAQ) and signal conditioning circuit composed of PC-DAQ test system; with GPIB, VXI, serial bus and field bus standard bus equipment such as the composition of the hardware approach GPIB systems, VXI system, a serial bus system, field bus systems. The current commonly used method is to insert a computer data acquisition card, with software generating instrument panel on the screen, using the software for signal analysis and processing and so on. This is exactly the core of this design. We use PCI bus technology has developed a virtual instrument system hardware platform, based on the PCI bus data acquisition cards, including signal conditioning and data acquisition of two parts, data acquisition card in combination with computer data-processing software can construct a variety of virtual instrument. Figure 1 constitutes a block diagram of the virtual instrument system.

LabVIEW development platform based on the design of the sound energy

Figure 1 constitutes a block diagram of the virtual instrument system

Speech signal time-domain approach

Digital speech processing methods are often divided into time-domain method, frequency domain method, with the state method, linear predictive coding methods and other methods. Time-domain approach involves the speech signal waveform, using time-domain method to analyze the voice signal characteristic parameters include the voice of the short-term average magnitude and energy, short-time average zero crossing rate, short-term autocorrelation function, as well as short-time average magnitude difference function and so on. Use of these characteristic parameters can be analyzed or processing of voice, such as the Voicing classification, pitch detection. Due to space limitations, here only the average level of short-time energy and short-term analysis. Speech signal time-domain analysis is the analysis and extraction of speech signals in time domain parameters, with the following features.

⑴ said the speech signal more intuitive and clear physical meaning;

⑵ realize a relatively simple operation less;

⑶ voice can be an important parameter;

⑷ the use of general-purpose oscilloscopes and other devices can see the change, easy to use.

Taking into account the above-mentioned advantages, this article on the speech signal time-domain approach for system analysis.

● Short-term Energy Analysis

Voice signal is a signal change over time, is voiced or voiceless incentive motivation, voiced the pitch cycle and the signal amplitude and so on change over time, this change is slow, and may be concluded that a short time, for example, 10 ~ 20ms in the same speech signal approximation. As a result, voice signals can be divided into a number of short-segment (also called analysis frames) to be processed. These short segment has a fixed nature, short paragraphs often have a certain degree of overlap between the composition of a voice. This method is called "short-term" approach. This short-term treatment of type 1 can be expressed.

LabVIEW development platform based on the design of the sound energy (1)

Where T [] expressed their voice to transform, this transformation is not necessarily linear, it can be nonlinear. The transformed sequence is multiplied by window function. This window sequence is located in line with the sample sign n the time, the width of the window function is limited, and then summing the product of all non-zero values obtained Qn, is n for the time T [x (k)] of the part of the weighted average. Short-term energy is usually defined as:

LabVIEW development platform based on the design of the sound energy (2)

This expression can also be expressed as:

LabVIEW development platform based on the design of the sound energy (3)

Figure 2 shows the corresponding mathematical model schematic.

LabVIEW development platform based on the design of the sound energy

Figure 2 Analysis of mathematical model of short-term energy

In summary, the voice short-term energy that depends on h (n) of the selection or w (n) option, usually using two windows, one of which a rectangular window, its window function, such as type 4 below.

LabVIEW development platform based on the design of the sound energy (4)

Experimental results show that, En was significantly less than the voiced segment of energy. Thus, according to short-term energy function can roughly distinguish between voiced and voiceless. En the case of high SNR, the use of short-term energy function can also distinguish between sound and silence.

● Short-term average rate of

Defined by the type of short-term energy function due to the need to calculate, for high signal to the square of its value after the later the greater, while for low-level signals whose value is less than a value, through the square even after the small . Resulting in short-term energy E (n) for the signal level value is too sensitive. In order to overcome this shortcoming, defines a short-term average rate function, see the type 5. The average energy of the mathematical model shown in Figure 3.

LabVIEW development platform based on the design of the sound energy

Figure 3 Analysis of mathematical model of the average energy

LabVIEW development platform based on the design of the sound energy (5)

Type instead of using the absolute value of the signal the square of the value of the signal. Mn can better reflect the magnitude of Mn changes within the voiceless. In this regard, Mn than En good, Mn can reflect the rate of change in dynamic range is also better than the En. Reflected in the voiceless voice and power adjustment between the voiced speech than En less obvious.

Short-term Short-term average rate of energy and the primary purpose is as follows.

(1) can distinguish voiced from the voice to, because the energy than the voiceless voiced a large number;

(2) can be used to distinguish between consonants and vowels of the boundaries, the boundaries of silent and sound, and even word boundaries, etc.;

(3) The segment information as a voice for speech recognition.

LabVIEW software design

Software design process can be divided into signal acquisition and processing processes. Flow waveform signal acquisition hardware, as shown in Figure 4, after collecting the signal analysis and processing flow shown in Figure 5, namely, the average energy and short-term energy analysis.

LabVIEW development platform based on the design of the sound energy

Figure 4 Waveform Acquisition Process

LabVIEW development platform based on the design of the sound energy

Figure 5 Analysis of the sound energy flow chart


Experiment, through the single-ear (ear plugs one end of the line should be exposed to the signal access) will be the music being played on the computer as an input signal, access to PCI-6025E data acquisition card, set the data acquisition cartoon Road to 0, select sampling point 50, the sampling rate of 8000, obtained experimental results shown in Figure 6. Select the computer randomly playing music as data entry, but did not use the function of an ideal sine wave signal generator waveform, because the regularity of the ideal sine wave is a periodic change, for the short-term or short-term average energy of the average range, The effect will not be so obvious, and thus easier for the data analysis. Can be seen from Figure 6, the original sound wave is -0.02 ~ 0.02, between the 0 ~ 100s period of time, with an average gradient of magnitude greater than the short-term energy and higher, they are concentrated in the 0 to 0.3 between the sound than the original waveform amplitude to be large, while the short-time energy is less than 0.005 the highest point of the range, 100s, the short-time energy and average magnitude of the trend is basically the same, tends to 0.

LabVIEW development platform based on the design of the sound energy

Figure 6 The results of the data and curves


In this paper, the short time of sound energy of LabVIEW and the average height analysis. Data acquisition card can be the input signal acquisition, signal segment as a voice message on the voice identification, and can distinguish voiced from voice to, but also can be used to distinguish between the boundaries of consonant and vowel, silent and there is the boundaries of sound, and even word boundaries and so on. To use LabVIEW to deal with the sound than the c programming language is more simple, and its beautiful interface, and the results can be directly described in the previous panel.

Digital Electronic Circuit Articles

Can't Find What You're Looking For?

Rating: Not yet rated


No comments posted.