Introduction
In the foreign language multimedia teaching, requests to the language fast to carry on the speed control, adapts the varying degree student’s demand. However, traditional pronunciation speed change product often while the teacher changes language fast, also changed the original speaker’s intonation, cannot achieve the teaching the true goal. Therefore, the pronunciation speed change system must have adjustment language fast at the same time, but also needs to guarantee that the original speaker intonation maintains the invariable characteristic. This article introduces is one kind based on the TMS320C5409 pronunciation real-time speed change system. Moreover, considered while the actual system the pronunciation compression memory and the pronunciation speed change often is the demand, therefore this article proposed that one kind based on the LPC low bit rate speech coding algorithm’s pronunciation speed change algorithm, this algorithm can adjust the pronunciation language willfully fast.
LPC algorithm
The LPC code regards as voice signal s(n) a time-variable sound track system under the glottis drive output signal. Speaking of the voiced sound, the glottis drive is one cyclical pulse train; But speaking of the voiceless sound, glottis drive for stochastic noise sequence. Voice signal’s simplified diagram as shown in Figure 1.
In the LPC code, may indicate a voice signal with the simplified model parameter that like the voiced sound, the voiceless sound distinguished, the tone cycle, increases G as well as digital filter coefficient {a1}, like this may obtain 3kb/s the code. When decoding, because in the LPC code, the voiced sound may regard as is one cyclical pulse train drive, the repetition interval is the tone cycle, therefore, to change the pronunciation the speed, we lengthen the pronunciation frame size, in joins certain pulse trains the drive then to obtain the speed change voice signal again.
Hardware system design
This system is a perfect application system, mainly completes to voice signal real-time LPC arranges the decoding and the pronunciation speed change, in addition, but also has the keyboard control function and the pronunciation recording and sending out function. Therefore, not only this system needs the exterior SRAM depositing pronunciation data, moreover needs the keyboard control device which connects with DSP. This system mainly by constitutions and so on TMS320C5409, TLC320AD50, SST39VF400A and CY7C133, system diagram as shown in Figure 2. The TMS320C5409 interior provides 32K character internal RAM uses for to deposit the real-time movement the procedure and the data, outside after the piece SRAM(CY7C133) uses in deposits the code the pronunciation data. System’s work process is as follows: After the system adds the electricity, constructs through the replacement causes TMS320C5409 to reposition. After TMS320C5409 replacement, solidifies from the interior commits suicide the master program Flash(SST39VF400A) on procedure and the initialization data removal to internal RAM, then the TMS320C5409 basis keyboard’s control execution operation, moves the LPC code operation (sound recording), the LPC decoding and the pronunciation speed change (playback), adjustment pronunciation speed change speed.
TMS320C5409 uses in realizing the pronunciation to arrange the decoding algorithm, TLC320AD50 uses in completing the voice signal the number/mold and the mold/number transformation function. Between TMS320C5409 and TLC320AD50 through McBSP serial port correspondence. In order to reduce CPU the burden, in this system designs McBSP receives or transmits one group of units after DMA, again to CPU interrupt way with the CPU correspondence. TMS320C5409 and TLC320AD50 connection as shown in Figure 3.
Uses DMA the way, namely the serial port receives or transmits one group of units every time, can autotrigger DMA move it delivers in internal Buffer, and so on Buffer filled told the CPU processing again through the interrupt mode. By now DMA should better use ABU (Auto Buffering) the pattern, may prevent in Buffer the data to compare Gao Shi effectively in the serial port speed the question which is flushed out by the recent data. When carries on the pronunciation speed change, only needs the 13MA channel’s frame size to do the corresponding change then.
SST39VF400A (256K×1 6) take TMS320C5409 the exterior data-carrier storage, the address bus and the data bus meets to the TMS320C5409 external bus, CR meets to TMS320C5409 DS. TMS320C5409 and SST39VF400A connection as shown in Figure 4.
What the exterior memory uses is CY7C133, it is high speed 2K×16bit static asynchronous twin port RAM, its storing velocity is 25ns. It has two sets of independent address wires, the data line and the control holding wire, allows in two control component’s data through the memory which connects together to carry on the correspondence. This twin port RAM allows two controllers simultaneously to read any memory cell (including simultaneously to read identical unit), as soon as soon as but does not allow simultaneously to write or reads writes the unified address unit.
Regarding TMS 320VC5402, the data-carrier storage CY7C133 corresponding address is 4000H~47FFH.
System software
This system realizes the goal is the keyboard control function and the pronunciation recording and sending out function, and can input voice signal by random variable speed broadcast, but does not change the pronunciation the intonation. The system uses the C language to carry on the function development.
In this system maps into the HPI mouth GPIO, and passes the keyboard segment ” or ” the gate receives INTO. When has the pressed key presses down, the INTO interrupt handling routine through the value judgment which reads out from GPIO this time is the sound recording, the playback, the language fast changes quickly the language fast slows down.
Antijamming design
In this system already uses the hardware antijamming technology, also has used the software antijamming technology. Below introduced briefly this system uses hardware watch-dog and digital filter’s design.
Sharpens system’s antijamming ability using the hardware watch-dog function. By the special-purpose chip MAX692 constitution’s watch-dog electric circuit (Figure 5), the system uses the periphery part to be few. MAX692 is the micro system monitoring circuit chip, has reserve functions and so on battery cut, power failure distinction, watch-dog monitoring. And WDI is the watch-dog examination input end, receives on a digital circuit system’s special-purpose I/O mouth or a main line mouth. RESET is the reset signal out-port, meets the digital circuit system’s replacement to carry RST. The MAX692 WDI fixed time cycle is 1.6 s, the reset pulse width is 200ms. If WDI maintains Gao Huodi surpasses ” the watch-dog ” fixed time the cyclical (1.6s), RESET end will have 200ms widely (smallest 140ms) negative pulse to cause the digital circuit system reset.
Supposes the digital filter’s input signal is X(n), the output signal is Y(n), then between the list entries and output sequence’s relations available difference equation expression is:
In the formula input signal X(n) may be the voice signal after the sampling and the ADC transformation obtains digital sequence, may also be computer’s output signal; aK and bK are
The coefficient, through establishes aK and bK may the bandpass filter which designs DF needs.
The digital filter’s software design method includes: Procedure judgment filtering (restriction filtering), position value filtering, arithmetic mean filtering, recursion average filtering, against pulse interference average filtering, first-order lag filtering and so on 6 methods. According to need, this system selection program judgment filtering.
Ties wu
The practice proved that this system while carries on the pronunciation speed change, can guarantee that the original speaker intonation maintains invariable, and the simplicity of operator, the system time delay is small.
