documentation - OHM ECCE - Ateneo de Manila University
Contents
1. file audiofrequencyanalyzerfrm h author Gerrald Mateo Keith Pulmano Mae Villanueva Created 9 18 2010 9 49 04 PM section DESCRIPTION audiofrequencyanalyzerfrm class declaration I I ifndef AUDIOFREOUENCYANALYZERFRM H define AUDIOFREOUENCYANALYZERFRM H ifdef BORLANDC pragma hdrstop endif ifndef WX PRECOMP include lt wx wx h gt include lt wx frame h gt else finclude lt wx wxprec h gt endif include lt wx dcclient h gt include lt wx colour h gt include xmcs h include utility h include lt complex gt Header Include Start include lt wx stattext h gt include lt wx textctrl h gt include lt wx button h gt Header Include End Dialog Style Start fundef audiofrequencyanalyzerfrm STYLE define audiofreguencyanalyzerfrm STYLE wxCAPTION wxRESIZ E BORDER wxSYSTEM MENU wxMINIMIZE BOX wxMAXIMIZE BOX wxCLOSE BOX Dialog Style End class audiofrequencyanalyzerfrm public wxFrame private DECLARE EVENT TABLE public audiofrequencyanalyzerfrm wxWindow parent wxWindowID id 1 const wxString title wxT Audio Frequency Analyzer const wxPoint amp pos wxDefaultPosition const wxSize size wxDefaultSize lon2. Event Table Start BEGIN EVENT TABLE audiofreguencyanalyzerfrm wxFrame EVT CLOSE audiofreguencyanalyzerfrm OnClose EVT BUTTON ID ANALYZEAUDIO audiofrequencyanalyzerfrm WxButton1Click END EVENT TABLE Event Table End audiofreguencyanalyzerfrm audiofreguencyanalyzerfrm wxWindow wxWindowID id const wxString amp title const wxSize size long style wxFrame parent id title position parent const wxPoint position size style graph new wxClientDC this CreateGUIControls audiofrequencyanalyzerfrm audiofrequencyanalyzerfrm void audiofrequencyanalyzerfrm CreateGUIControls 37 GUI Items Creation Start numberOfAudioBlocks new wxStaticText this ID NUMBEROFAUDIOBLOCKS wxT Number of Audio Blocks wxPoint 222 16 wxDefaultSize 0 wxT numberOfAudioBlocks countAudioBlocks new wxTextCtrl this ID AUDIOBLOCKS wxT wxPoint 366 14 wxSize 87 22 0 wxDefaultValidator wxT countAudioBlocks countAudioBlocks gt Enable false analyzeAudio new wxButton this ID ANALYZEAUDIO wxT Analyze Audio wxPoint 488 10 wxSize 99 29 O wxDefaultValidator wxT analyzeAudio SetTitle wxT Audio Frequency Analyzer SetIcon wxNullIcon SetSize 8 8 783 240 Center E fa o GUI Items Creation E void audiofrequencyanalyzerfrm OnClo
3. list lt char gt iterator bli ref for bli audioBuffer begin bli audioBuffer end bli ref bli if ref audioBuffer end block size lastBufferSize int bufferSize block size const char buffer bli for int i 0 i lt bufferSize i 2 short int numl6 numl6 unsigned char buffer i lt lt 8 unsigned char buffer i 1 dataList push back numl6 list that contains integer audio blocks bool XMCS Play const int BLOCK COUNT 10 HWAVEOUT hWaveOut WAVEHDR waveBlocks int waveCurrentBlock int block size char buffer INITIALIZATION AND MEMO block size maxBufferSize waveBlocks new WAVEHDR B for int i 0 i lt BLOCK COUNT memset amp waveBlocks i O waveBlocks i lpData freeBlockCount BLOCK COUNT waveCurrentBlock 0 Initializec 52 number of audio blocks for playback handle array of wave headers index of current block being played size of block to be played temporary buffer RY ALLOCATION OCK_COUNT i sizeof WAVEHDR new char block size riticalSection amp criticalSection PLAYBACK open audio output device if waveOutOpen amp hWaveOut WAVE MAPPER amp wfx DWORD PTR waveOutProc DWORD PTR this CALLBACK FUNCTION l MMSYSERR_NOERROR return false list lt char gt iterator bli a
4. 4000Hz Each one was saved as a wav file and played in a silent room while the audio freguency analyzer was running Also the volume of the playback was adjusted during each tone playback such that noise is as less noticeable as possible This experiment was done so that the analyzer may be checked if it displays correct peaks at different frequencies The results of the tests are as follows Fig 6 4 1 Analysis of a 500 Hz sinusoidal waveform Fig 6 4 2 Analysis of a 1000 Hz sinusoidal waveform 28 Fig 6 4 3 Analysis of a 1500 Hz sinusoidal waveform Fig 6 4 4 Analysis of a 2000 Hz sinusoidal waveform 29 Fig 6 4 6 Analysis of a 3000 Hz sinusoidal waveform 30 Fig 6 4 8 Analysis of a 4000 Hz sinusoidal waveform 31 6 5 Monophonic Polyphonic and True Tones To further test the accuracy of the audio freguency analyzer we varied the input audio from monophonic tones to polyphonic tones and finally to true tones The monophonic tone was generated by playing a recorded Nokia 3310 ringing tone The polyphonic tone was generated by a midi audio file which was played from a Sony Ericsson R300 cellular phone And finally the true tone was generated by an mp3 file of the song Imagine by the Glee Cast We used these three types of tones because their freguency spectrum should noticeably vary from one type to another A monophonic tone is expected to have the least frequency components since it only uses o
5. 8 0 for int i mark i lt bin length i num num 2 bin i 1 1 0 if not 1 then assign 0 for int i 0 i lt mark i num num 2 bin i 1 1 0 if not 1 then assign 0 return num bool Utility IsBinaryDigit char digit return digit 0 digit 1 bool Utility IsDecimal string sequence for int i 0 i lt seguence length i if isdigit sequence i return false return true bool Utility IsHex string sequence for int i 0 i lt seguence length i if isxdigit sequence i return false DET true bool Utility IsBinary string sequence for int i 0 i lt seguence length i if seguence i 0 sequence i 1 return false return true 63 A 1 8 utility h ifndef UTILITY HO define UTILITY H_ include lt string gt using std string class Utility private returns the hexadecimal symbol representing the integer if the integer is less than 16 0 if not wA static char Symbol int public converts a 16 bit integer into its hexadecimal string little endian representation that is made up of a specified number of bits 8 or 16 kaya static string Intl ToHexString int integer int bits converts a 16 bit integer into its binary string little endian representation that is made up
6. buffer INITIALIZATION isFirst true isStopped false maxBufferSize bufferSize 2048 currentSet 0 RECORDING open audio input device and allocate memory for recorded samples do for 2 sets also set values of wave header fields for nt 1 SOF 1 lt 2 144 open device if waveInOpen shWaveIn i WAVE MAPPER wfx 0 0 CALLBACK NULL l MMSYSERR_NOERROR return false T allocate memory waveHeader i lpData new char bufferSize initialize waveHeader i dwBufferLength bufferSize waveHeader i dwFlags 0 waveHeader i dwLoops 0 prepare a wave header for the current set to be used for recording if wavelnPrepareHeader hWaveln currentSet waveHeader currentSet sizeof WAVEHDR MMSYSERR NOERROR goto failed cleanup 55 add buffer for the current set to be used for recording if wavelnAddBuffer hWaveln currentSet amp waveHeader currentSet sizeof WAVEHDR MMSYSERR_NOERROR goto failed cleanup start recording if wavelnStart hWaveIn currentSet MMSYSERR_NOERROR goto failed cleanup buffer new char bufferSize for buffer currently being recorded otherSet I currentSet currentSet 1 2 for the first iteration only statements inside are not executed if lisFirst memcpy buffer LPSTR waveHeader otherSet lpDa
7. buffer i nextSample amp OxFF00 gt gt 8 readBytes fin eof i 1 bufferSize if fin eof break break default return 0 return readBytes 47 int XMCS WriteBlockFromBuffer ofstream amp fout int bufferSize const char buffer int channels GetChannels int bits GetBits if bufferSize 2 bufferSize lt 0 return 0 const int samplesPerLine channels 2 fout lt lt n switch bits case 8 for int i 0 i lt bufferSize i if i samplesPerLine fout lt lt n if radix 2 fout lt lt Utility Intl 6ToBinaryString buffer i SJT Wa else if radix 16 fout lt lt Utility Int16ToHexString buffer i 8 lt lt us break Case 16 short int numl6 for int i 0 i lt bufferSize 1 2 numl6 unsigned char buffer i lt lt 8 unsigned char buffer i 1 if 1 2 5 samplesPerLine fout lt lt n if radix 2 fout lt lt numl6 lt lt endl Utility Intl6ToBinaryString numl6 16 lt lt else if radix 16 fout lt lt Utility Int16ToHexString num16 16 lt SU WA break default return 0 return bufferSize 48 short int XMCS GetNextSample ifstream fin int bits GetBits char temp string sample sample clear switch radix case 2 for int i 0 i lt bits i temp GetNextNonWhiteSpace fin an un
8. free audio block while freeBlockCount if kbhit getch waveOutReset hWaveOut bli audioBuffer end break current block is next block imagine array is circular waveCurrentBlock waveCurrentBlock BLOCK COUNT unless playback is stopped wait for all blocks to complete while freeBlockCount lt BLOCK COUNT if kbhit getch waveOutReset hWaveOut break CLEAN UP unprepare any blocks that are still prepared for int i 0 i lt freeBlockCount i if waveBlocks i dwFlags amp WHDR PREPARED waveOutUnprepareHeader hWaveOut waveBlocks i sizeof WAVEHDR DeleteCriticalSection amp criticalSection waveOutClose hWaveOut delete buffer for int i 0 i lt BLOCK COUNT i delete waveBlocks i lpData delete waveBlocks return true 54 bool XMCS Record Unload clear list HWAVEIN hWaveln 2 handles WAVEHDR waveHeader 2 wave headers bool isFirst 1 indicates that first buffer is being filled bool isStopped 1 indicates recording was stopped by a key press char buffer temporary buffer int bufferSize recording buffer size int currentSet set of handle and wave header being used int otherSet set of handle and wave header being prepared int lastByteCount number of recorded bytes on the last
9. function play function and save function A new function is also added to this class It is used to access the audio blocks to be able to transfer them for FFT computation later on The changes made to the original files are also indicated in this chapter Some minor changes are also done to the record and save functions for maximized compatibility with the graphical user interface and checking Parts that were not directly used or altered will not be discussed anymore because it is already expounded in the XMCS Player Recorder documentation Baterina and Macalintal 2008 2 1 XMCS Class This class is used because it has audio recording and playing capabilities which are needed for the audio freguency analyzer Furthermore it is able to provide direct access to the audio blocks being recorded into memory which are needed for the FFT computation The audio handling parts of the analyzer are taken care of by this class 2 1 1 XMCSO Constructor This constructor is the one that sets the default values to be used if the user does not opt to change the settings of the audio format For the frequency analyzer the user is not asked to set these settings so the default values indicated here are the ones used throughout the project Important parameters are bits per sample and samples per second These are set to 16 and 8000 respectively 2 1 2 XMCS Record function This function is a vital part of this project because it is the function that st
10. gt f 2m 1 W 3 2 1 F k G k W H k 3 23 By continuously dividing up or decimating the sample sequence into odd and even sub sequences we can turn the complex DFT equation into a set of 1 point DFT s such that our N point DFT becomes a combination of these smaller and simpler DFT s This is the goal of the decimation in time algorithm Note however that the decimation only works 1f the number of our sampling points is a power of 2 Meanwhile the following properties are useful to further simplify the DFT equation Periodicity Property Wy W 3 3 1 k N 2 gt Symmetry Property W y W 3 3 2 The symmetry property is used to obtain a function as shown in eq 4 F k N 2 G k W H k 3 4 12 Thus the DFT of next half of the N samples can be obtained by using the same set of computations from the the first half and then just changing the operation to subtraction For illustrative purposes the figure below shows the computation of an 8 point DFT ee caltech edu 2010 i Stage 1 Stage 2 Stage 3 x 0 X 0 x 4 xa x 2 X 2 x 6 XG x 1 X 4 x 5 X 5 x 3 X 6 x 7 xm Figure 3 1 Eight point decimation in time FFT algorithm tt Aa O We tes LEST Figure 3 2 Basic butterfly computation in the decimation in time FFT algorithm 13 3 1 2 Bit Reversal Since the decimation in time algorithm involves rearranging our sample seguence it needs another algorithm that should put the samp
11. loop where the DFT computation is done complex lt double gt temp double m 6 28318530717959 k n np double q cos m double r sin m complex lt double gt W q r temp data n W F k F k temp increments for every loop to implement the summation process W 0 0 24 While the GUT is still being developed results generated from the test code and the FFT function are first displayed in the command prompt The number of inputs that it can process is however limited to 16 points only where above this number the window does not respond anymore The codes are tried out using 8 point and 16 point input samples The figure below shows the comparison for a 16 point sample seguence with inputs 0 to15 a DAUsers Xeith Docunents 3th year ist sen VILT 132MT A DAL serdKer hU x urnen SVOR por Tal sem FLO 157 WM M2 AP AP M 4d M a a B gt 236bbe UL3 gt ar 6913 gt Mpa e ES big 5 J4549 gt 8 gt 11 972R gt 19 3437 49 2187 had Vand xd br baz Laz bed Ps Per AAA A fel jm mm brk hk Jl MN CE m7 ON ST CN ek S P p RS had bm et kon n bond mnt bn bom bm k kl ll ET ni d w U any key to continue any key Lu vuntinus Figure 6 2 1 FFT code generated results Figure 6 2 2 Test code generated results for a 16 point sample for a 16 point sample It can be seen from the two figures that the results generated are the same except for one poin
12. of a specified number of bits 8 or 16 static string Intl6ToBinaryString int integer int bits converts a hexadecimal string little endian representation of a 16 bit integer into its numerical eguivalent static short int HexStringToIntl6 string hexString converts a binary string little endian representation of a 16 bit integer into its numerical equivalent 8 static short int BinaryStringToIntl6 string binString checks if the character is a binary digit static bool IsBinaryDigit char digit 64 checks if the string represents a decimal number A static bool IsDecimal string decString checks if the string represents a hexadecimal number static bool IsHex string hexString checks if the string represents a binary number x static bool IsBinary string binString endif A 1 9 fourier h include lt iostream gt include lt complex gt using namespace std class Fourier complex lt double gt data array 1024 int block length public void set int int void magnitude double void FFT function void private void bit reverse function void hep void Fourier set int data int length delete data_array block length length for int i 0 i lt block length i data array i complex lt double gt data i 0 65 void Fourier bit reverse
13. saves the audio samples into a file The data representation of the samples can either be binary of hexadecimal but for the analyzer 1t was used to save decimals bool Play This is a public function that plays the audio recorded from memory or audio from an XMCS file bool Record This is a public function that records audio and stores it into buffers and later into a linked list Any new recording overwrites the audio currently stored void AccessBlocks This is a public member used to access the audio blocks as integers for the FFT 10 2 2 Utility Class This class was not used for the later part of the analyzer development The primary function of this class is to convert integers into binary or hexadecimal data types It was used at first when the audio blocks were being accessed through save and WriteBlockFromBuffer functions However when the recording part of the analyzer was integrated with the FFT the audio samples were left as decimals so this class is unused in the final implementation 11 3 Fourier Class 3 1 Theoretical Background 3 1 1 Decimation in Time Algorithm Given below is the eguation for the Discrete Fourier Transform Fins gt yle Y K 3 1 nk Jy K is referred to as the twiddle factor Wy e where We can further break up this equation by grouping together its even and odd components as shown in the following equation FUE 2 flaw 3 2 1 Flkl X fl2m1W W
14. syntax int ctr 0 juas de countAudioBlocks gt Clear countAudioBlocks lt lt ctr The first line is placed before the while loop That denotes where the counting should start and it ends with the second line placed right after the loop which graphs the FFT This is where the counts starts at 1 and so on The fourth line shows the number on the output field The third line clears the previous number so that the output field shows increments of integer 19 5 User s Manual 5 1 CDROM Contents lt CD root director yA index html homepage of project website software Output MingW audiofrequencyanalyzer exe Audio Frequency Analyzer source_code xmes h XMCS class definition xmcs cpp XMCS class implementation utility h Utility class definition utility cpp Utility class implementation fourier h Fourier class definition audiofrequencyanalyzerfrm h audiofrequencyanalyzerfrm class definition audiofrequencyanalyzerfrm cpp audiofrequencyanalyzerfrm class implementation audiofrequencyanalyzer dev Project compilation w all classes included audiofrequencyanalyzerapp h audiofreguencyanalyzerapp definition audiofreguencyanalyzerapp cpp audiofrequencyanalyzer main program doc documentation odt project documentation odt file documentation pdf project documentation pdf file 20 progress reports contains progress reports of group members mateo pulmano villanu
15. Audio Freguency Analyzer A Project by Gerrald Carlo A Mateo Zarah Katrina E Pulmano Maria Mae Fleur M Villanueva Submitted to Luisito L Agustin Instructor ELC 152 In Partial Fulfillment of the Reguirements for the Course ELC 152 Signal Processing Department of Electronics Computer and Communications Engineering School of Science and Engineering Loyola Schools Ateneo de Manila University Ouezon City Philippines October 2010 Abstract An Audio Freguency Analyzer that displays an audio sample in the freguency spectrum is implemented using C First it is programmed to record 16 bit audio samples with a sampling rate of 8 kHz Another class is created to provide functions that get the Discrete Fourier Transform DFT of the audio samples Finally a graphical user interface is developed to display the freguency spectrum obtained from the DFT computations Note that the recording and graphing of the freguency spectrum is done continuously and by blocks where each audio block consists of 1024 samples Acknowledgments Our sincerest acknowledgement to the following people who contributed to the success of this project First we would like to thank our professor Mr Luisito Agustin for allowing us to develop an understanding of our project for always making himself available for consultations and for directing us towards the accomplishment of this project Special thanks goes to our classmates Mr Ricson Chua a
16. arts the whole process The samples used for the FFT computation is taken from output of the recorder Therefore the output of the recorder is the input of the FFT For this project the audio block size for recording is set to 1024 a power of 2 samples so that it can be fully accommodated and processed by the FFT function Details on the FFT will be discussed in chapter 3 Audio is recorded directly from an audio input device and is stored into temporary buffers and then into a linked list audioBuffer To be able to do this the audio input device is first opened for recording After which the buffers that will be filled with the audio samples from the input device are prepared Two temporary buffers are used so that there is always an available buffer to store the data once one of the buffers is filled If a buffer is filled the contents are automatically transferred to the end of audioBuffer After it is successfully copied the buffer is immediately cleared for the next set of audio samples Originally the record function was intended for console applications This made it possible to interrupt recording at any time However since the audio frequency analyzer is a windows application the recording cannot be interrupted without the use of threads To address this problem the recording function was altered to stop after reaching 1024 samples In the process the buffer size was changed to 1024 samples with 16 bits per sample or simply 2048 byt
17. boost set to 30dB in a quiet room By testing the software several times we concluded that the optimal microphone volume is 30 because it lowers the noise considerably and it makes the program more reactive to the audio played within the range of the microphone The resulting graph in a silent room with volume at 30 is shown in Figure 5 3 4 Fig 5 3 4 Analyzed output with microphone volume set to 30 23 6 Tests Following are tests done to verify the accuracy of the Audio Freguency Analyzer Project 6 1 LabVIEW For comparison purposes a simple FFT system was developed using the LabVIEW environment where the input is an 8 point sample with values from 0 to 7 The DFT computation results that were generated from the LabVIEW are then compared to those generated from the FFT decimation in time function code that was created Note that the input sample seguence for the FFT function is just the same as our input in the Labview As observed both produced similar results 6 2 Decimation in Time FFT function vs DFT Test Code Another test code which is patterned to the original DFT eguation is also created to further check the accuracy of our decimation in time FFT function We verified if the results generated using this code matches those that were obtained using the FFT function The code is shown below for int k 0 k lt np k loop to get the DFT for all points from 0 to np for int n 0 n lt np n
18. corded using printf or cout The audio blocks that were displayed were then compared with the ones in the files generated by the save function that was also included in the code right after recording This gave us the assurance that we are indeed accessing the correct audio samples from the buffers 35 7 Results and Recommendations A simple audio freguency analyzer with its basic function has been successfully implemented Its main features are as follows support for 16 bit audio flexible duration for recording To further improve the project it is recommended that the following features be added use of a range of sampling rates for the XMCS files pause and resume and stop operation during the recording by using threads and better resolution of the graph of the freguency spectrum and added options that will allow the user to record and graph in different sampling rates 36 A 1 Source Files Appendix Source Code A 1 1 audiofrequencyanalyzerfrm cpp E da dei SaaS file audiofrequencyanalyzerfrm cpp author Gerrald Mateo Keith Pulmano Mae Villanueva Created 9 18 2010 9 49 04 PM section DESCRIPTION audiofrequencyanalyzerfrm class implementation RE o o O ee AS eas include audiofrequencyanalyzerfrm h include xmcs h include utility h include fourier h sm en noo audiofrequencyanalyzerfrm VE ler RS ae EE sss
19. e new function in this class They were combined to create the AccessBlocks function 2 1 5 XMCS AccessBlocks function This is a new function added to XMCS to be able to transfer the audio blocks into a global linked list of integers By doing so the audio samples are easily accessed by the FFT function for computations The function is used to traverse the audioBuffer contents after recording While traversing audioBuffer the audio blocks stored in it are being stored into another linked list dataList The new linked list is a list of integers so it is easier to access compared to a list of pointers The structure of this function is basically the combination of the private function WriteBlockFromBuffer and the save function 2 1 6 Members In this section the key members used for the analyzer are described Old and new members of the XMCS class are included in this section list lt char gt audioBuffer This is a private member which is used to store the contents of the temporary buffers list lt int gt dataList This is a public member which stores the audio blocks and makes it accessible to the FFT function and the GUI cplusplus com 2010 list lt int gt iterator it This is a public member used to traverse the global list to be able to convert the list of integers into arrays for FFT It is used in the GUI for traversing the currently recorded audio samples bool Save string filename This is a public function that
20. eded for the first for loop which has the linked list and immediately the FFT is computed by taking the data from the second for loop that is the list provided by the XMCS class 18 compute set data np compute FFT function compute magnitude out The code above is setting the data to be computed and only the magnitude from the computed value are taken Immediately after the computation the graph is being shown First the x and y axes are drawn for reference purposes and then the values of the magnitudes of the FFT are shown which is inside the third for loop x 5 1 1 3 y 650 out i 2 graph gt DrawLine x 650 x y graph gt DrawCircle x y 2 In order to successfully loop the whole process the source code is consisted of lines that clears the record compute and draw functions Inside the XMCS class the Unload function is able to remove the previous recorded sound once a new sound is recorded Inside this source code the audio dataList clear clears the contents of the list before another data comes in the graph gt Clear is prompted to clear the background immediately before it draws the x and y axes and the magnitudes of the FFT The labels for 0 1000 2000 3000 and 4000 Hertz and the Magnitude are printed on the GUI using graph gt DrawText the text x coordinate y coordinate Counting the number of blocks processed is possible with the counter inside the source code with this
21. else return false Decimal info t samplesPerSec atoi info c str if t samplesPerSec 8000 t samplesPerSec 11025 t samplesPerSec 22050 t samplesPerSec 44100 if t samplesPerSec lt 8000 t samplesPerSec gt 44100 return fal return false fin gt gt keyword if keyword formatend return false LSC SetChannels t channels SetBits t bits SetSamplesPerSecond t samplesPerSec SetFileRadix t radix return true int channels int bits int samplesPerSec fout lt lt channels teFileHeader ofstream fout GetChannels GetBits lt lt channels lt lt nbits GetSamplesPerSecond lt lt bits lt lt nradix lt lt radix lt lt nformat lt lt samplesPerSec lt lt formatend n 46 if fout fail return false return true int XMCS ReadBlockToBuffer ifstream fin int bufferSize char buffer int bits GetBits if bufferSize 2 bufferSize lt 0 return 1 int readBytes 0 if fin eof readBytes 0 return readBytes switch bits case 8 for int i 0 i lt bufferSize i buffer i GetNextSample fin readBytes fin eof i 1 bufferSize if fin eof break break Case 16 for int i 0 i lt bufferSize 1 2 short int nextSample GetNextSample fin buffer i 1 nextSample OxFF
22. es Once the recording is halted the buffer contents are transferred to audioBuffer and the audio input device is closed A new recording will automatically overwrite the buffer contents from the previous recording 2 1 3 XMCS Play function The play function is not featured in the final executable application of the analyzer However it is important for the development of the project because it is used to check if the recorder successfully recorded the audio samples It is also useful in checking 1f the buffer size was properly allocated by the recorder The play function is capable of playing from an XMCS file or from previously recorded data However only playing from memory is utilized in the analyzer development Essentially in the play function the audio blocks taken from the linked list is played through an audio output device Several buffers are also used to avoid losing the data that are being extracted from audioBuffer 2 1 4 XMCS Save and WriteBlockFromBuffer function The save function is the primary function used to check if the data accessed from audioBuffer correct It is used to directly access the contents of audioBuffer It makes use of a private function the WriteBlockFromBuffer function to organize the format of the file It was tweaked such that instead of saving the samples as binary or hexadecimal the samples are saved into a file as signed decimal numbers These two functions are crucial in the development of th
23. eva proposal contains approved project proposal proposal odt 5 2 Software Features The XMCS Player Recorder allows the user to do the following Record audio Automatically display the frequency spectrum upon recording Modify the number of audio blocks to be recorded and display 5 3 Using the Software 5 bE a rae w E Fig 5 3 1 Opening Screen When the program is ran an opening screen Figure 5 3 1 is displayed The user can choose to maximize the window to view the graph in a better resolution 21 The user may then press the button Analyze Audio to start the recording and see a dynamic graph of the freguency spectrum of the audio recorded from the microphone It is also important to note that the settings of the microphone being used is crucial in obtaining meaningful and accurate analysis Having a very high microphone input volume would make the graph very noisy even if the surrounding environment is guiet as shown in Figure 5 3 2 Fig 5 3 2 Analyzed output with microphone volume set to 100 in a guiet room Moreover boosting the microphone input will also make the resulting graph inaccurate It is advisable to set the microphone boost setting to OdB because setting it to 30dB maximum boost with Windows 7 operating system will make the graph consistently noisy as well This result is shown in Figure 5 3 3 22 Fig 5 3 3 Analyzed output with microphone volume set to 100 and microphone
24. expected character means start of comment upto end of line while Utility IsBinaryDigit temp DiscardUpToEol fin temp GetNextNonWhiteSpace fin if fin eof break sample temp return Utility BinaryStringToInt16 sample break case 16 for int i 0 i lt bits 4 i temp GetNextNonWhiteSpace fin an unexpected character means start of comment upto end of line while isxdigit temp DiscardUpToEol fin temp GetNextNonWhiteSpace fin if fin eof break sample temp break return 0 return Utility HexStringToInt16 sample 49 char XMCS GetNextNonWhiteSpace ifstream fin char digit do if lineCharsRead 256 DiscardUpToEol fin digit fin get lineCharsRead digit n 0 lineCharsRead 1 while isspace digit amp amp fin eof if fin eof return 0 return digit void XMCS DiscardUpToEol ifstream fin char c do if fin eof break c fin get while c n lineCharsRead 0 RR RR KK RK k kk k kk k kk k kk k kk OK AA XMCS CLASS SPECIAL FUNCTIONS Jf KKKKK RR KKK KK KK k kk k kk OR AA bool XMCS Load string filename T const int BLOCK SIZE 1024 char tempBuffer char buffer ifstream fin fin open filename c str if fin fail return false if IReadF
25. function int index 0 k 0 complex lt double gt temp for int i 0 i lt block length i if index gt i Swapping of the data contents temp data array index data array index data arrayl il data array i temp k block length 2 while index gt k index index k k b s if k index 0 break index index k void Fourier FFT function call bit reverse function bit reverse function indices for the DFT int k 0 index 0 int mmax 1 istep sign 1 parameters for the computation of the twiddle factor double theta 0 0 ur 0 0 ui 0 0 wcos 0 0 wsin 0 0 while block length gt mmax istep 2 mmax theta 3 141592654 float mmax ur IO UL 0 05 wcos cos theta wsin float sign sin theta for int j 0 j lt mmax j declaration of the twiddle factor in complex form complex lt double gt twiddle factor ur ui for k j k lt block length k istep index k mmax complex lt double gt twiddleTemp data array index twiddle factor 66 calculation of the DFT s data_array index data_array k twiddleTemp data arraylk data_array k twiddleTemp calculation for the twiddle factor double wtemp wtemp ur wcos ui wsin ui ui wcos ur wsin ur wtemp mmax istep void Fourier magnitude double mag scaled co
26. g style audiofreguencyanalyzerfrm STYLE 40 virtual audiofrequencyanalyzerfrm XMCS audio void WxButtonlClick wxCommandEvent amp event private GUI Control Declaration Start wxClientDC graph wxTextCtrl countAudioBlocks wxButton analyzeAudio wxStaticText numberOfAudioBlocks private enum GUI Enum Control ID Start ID NUMBEROFAUDIOBLOCKS 1009 ID AUDIOB OCKS 1008 ID ANA YZEAUDIO 1006 GUI Enum Control ID End ID DUMMY VALUE private void OnClose wxCloseEvent amp event void CreateGUIControls eg endif A 1 3 audiofreguencyanalyzerapp cpp Name audiofreguencyanalyzerfrm cpp Author Gerrald Mateo Keith Pulmano Mae Villanueva Created 9 18 2010 9 49 03 PM Description include audiofreguencyanalyzerapp h include audiofreguencyanalyzerfrm h IMPLEMENT APP audiofrequencyanalyzerfrmapp bool audiofreguencyanalyzerfrmapp OnInit audiofrequencyanalyzerfrm frame new audiofrequencyanalyzerfrm NULL 41 SetTopWindow frame frame gt Show return true int audiofrequencyanalyzerfrmapp OnExit return 0 A 1 4 audiofreguencyanalyzerapp h Name audiofreguencyanalyzerapp h Author Gerrald Mateo Keith Pulmano Mae Villanueva Created 9 18 2010 9 49 03 PM Description ifndef AUDIOFREQUENCYANALYZERFRMApp
27. h define AUDIOFREQUENCYANALYZERFRMApp h _ ifdef BORLANDC pragma hdrstop endif ifndef WX PRECOMP include lt wx wx h gt else include lt wx wxprec h gt endif class audiofreguencyanalyzerfrmapp public wxApp public bool OnInit int OnExit v endif 42 A 1 5 xmcs cpp include xmcs h tinclude lt stdio h gt include lt iostream gt include lt complex gt using namespace std FUNCTION DECLARATIONS iii void bit reverse f za XMCS C unction complex lt double gt a void FFT function complex lt double gt a int length RR KR KR KKK k kk k kk k kk kk k kk kk KX KA KA KA X k X k k kk k LASS CONSTRUCTORS int leng AND th KOR KKK KKK KKK KKK DESTRUCTOR Ky RR KR KKK KKK KK KR KK k kk k kk k kk k RK k kk kk OK RK I X KA X OR KK ORK KK k XMCS XMCS audio format wfx nChannels T wfx wBitsPerSample 16 wfx nSamplesPerSec 8000 wfx cbSize 0 wfx wFormatTag WAVE FORMAT PCM wfx nBlockAlign wfx wBitsPerSample wfx nAvgBytesPerSec wfx nBlockAlign wfx others radix 2 maxBufferSize 0 lastBufferSize 0 XMCS XMCS XMCS amp xmcs wfx xmcs wfx radix xmcs radix maxBufferSize xmcs maxBufferSize lastBufferSize xmcs lastBufferSize int block size maxBufferSize char buffer list lt char gt iterator bli bli xmcs aud
28. he block containing the input for the FFT computation and the length of the data block or the number of sampling points that will be computed Before running or calling the other functions in this class it is necessary that this function is called first void FFT function void This function is called to get the Fast Fourier transform of the complex data array that is set using the set function void bit reverse function void This function performs the bit reverse function that is needed before performing the main FFT computation Note however that there is no need to call this function anymore because it is automatically performed when the FFT function is called void magnitude double mag This function computes and stores into an array the magnitudes of each point in the FFT computation It requires a parameter which is a double type array that will be used as the storage for the magnitude computations 15 4 User Interface Implementation The user interface of the Audio Freguency Analyzer program is designed using the wx Widgets cross platform C GUI toolkit Koga 2009 The graphical user interface of the program allows to analyze an audio recorded from the sound card and count the number of audio blocks being processed The window of the GUI is composed of the output field the control button and the graph field as shown Fig 4 Fig 4 Audio Frequency Analyzer GUI The source code involved in the GUI are
29. i for bli ref blo ref If xmcs audioBuffer begin oBuffer r xmc ck size ref bli end bli s audioBuffer end lastBufferSize buffer memcpy bu audioBu new char block size ffer bli block size ffer insert audioBuffer end 43 wfx nChannels gt gt 3 nSamplesPerSec buffer XMCS XMCS Unload Jf KKKKK KKK k kk KK RK KK k kk k KR KK k AA XMCS CLASS ACCESSOR FUNCTIONS av RRR RK KKK KKK KKK KKK KKK KKK KKK KKK KKK KKK KKK KKK KKK KKK KKK KKK KKK KKK KRKKKKK bool XMCS SetChannels int val if val val 2 return false wfx nChannels val wfx nBlockAlign wfx wBitsPerSample wfx nChannels gt gt 3 wfx nAvgBytesPerSec wfx nBlockAlign wfx nSamplesPerSec return true bool XMCS SetBits int val if val val 16 return false wfx wBitsPerSample val wfx nBlockAlign wfx wBitsPerSample wfx nChannels gt gt 3 wfx nAvgBytesPerSec wfx nBlockAlign wfx nSamplesPerSec return true bool XMCS SetSamplesPerSecond int val if I val 8000 val 11025 val 22050 val 44100 if val lt 8000 val gt 44100 return false wfx nSamplesPerSec val wfx nBlockAlign wfx wBitsPerSample wfx nChannels gt gt 3 wfx nAvgBytesPerSec wfx nBlockAlign wfx nSamplesPerSec return true bool XMCS SetFileRadix
30. ileHeader fin return false Unload tempBuffer new char BLOCK SIZI lineCharsRead 0 Gl i s 50 buffer new char BLOCK_SIZE lastBufferSize ReadBlockToBuffer fin BLOCK SIZI tempBuffer memcpy buffer tempBuffer BLOCK SIZE audioBuffer insert audioBuffer end buffer while lastBufferSize BLOCK SIZE El maxBufferSize BLOCK SIZI El delete tempBuffer fin close bool XMCS Unload if laudioBuffer size return false int i audioBuffer size int j 0 while audioBuffer size 66 maxBufferSize if audioBuffer size 1 lastBufferSize delete audioBuffer begin audioBuffer erase audioBuffer begin lastBufferSize maxBufferSize 0 return true bool XMCS Save string filename int block size maxBufferSize ofstream fout fout open filename c_str if fout fail return false if IWriteFileHeader fout return false list lt char gt iterator bli ref for bli audioBuffer begin bli audioBuffer end bli ref bli if ref audioBuffer end block size lastBufferSize WriteBlockFromBuffer fout block size bli 51 fout close return true accesses audioBuffer and transfers void XMCS AccessBlocks into integer list int block size maxBufferSize
31. in Figure 6 6 1 Fig 6 6 1 Screen shot of the timer and the number of audio blocks processed 6 7 Recorder Testing The recorder part of this project was tested using the play and save functions of the XMCS player recorder Before implementing the recorder on the GUI we first tested the function in a console application At first we created a program that records and then plays the recorded audio right away A keyboard hit will trigger the start and end of the recording process The next keyboard hit will play the recorded audio This test makes sure that recording does take place with the record function Since the final output of this project is a windows application we excluded the 34 keyboard hit part which is only applicable to console applications To stop the recording process we replaced the keyboard hit function with a loop that will stop when the audio block size is full The audio block size was also set to 1024 samples or 2048 bytes for 16bits per sample After doing so we tested the program by playing the audio after recording We expected short audio lengths to be played depending on the audio block size that we set This test was also successful Finally we had to access the audio blocks that were recorded so we used the save function since this is the part of the original code that directly accesses the audio samples in the memory buffers First we created a program that would display the audio blocks that were re
32. ing std ofstream using std list class XMCS private to store audio format ES WAVEFORMATEX wfx list for audio samples and buffer size indicators la list lt char gt audioBuffer int maxBufferSize int lastBufferSize 58 file radix needed for reading and writing Ai int radix file input functions e char GetNextNonWhiteSpace ifstreame void DiscardUpToEol ifstream short int GetNextSample ifstream amp bool ReadFileHeader ifstream int ReadBlockToBuffer ifstream amp int char file output functions El bool WriteFileHeader ofstream amp int WriteBlockFromBuffer ofstream int const char others shared by functions like global variables E short int lineCharsRead CRITICAL SECTION criticalSection volatile int freeBlockCount public constructors destructor Kol XMCS XMCS XMCS amp XMCS list lt int gt dataList list lt int gt iterator it load audio samples from a file filename to the list audioBuffer E bool Load string filename empty the list audioBuffer K bool Unload save audio samples in the list audioBuffer to a file x filename 7 bool Save string filename 59 play audio samples in the list audioBuffer record audio samples and store to the l
33. int val if val 2 val 16 return false radix val return true 44 int XMCS GetChannels return wfx nChannels XMCS GetBits return wfx wBitsPerSample XMCS GetSamplesPerSecond return wfx nSamplesPerSec XMCS GetFileRadix return radix RR KR KKK KK kk kk kk k kk k AA ja XMCS CLASS FILE I O OPERATIONS S Jf KKKKK kok k k kk k kk kk k kk k kk k kk k kk k kk k kk OR K KA A kA X KA X KA AX AKA A k k k kk bool XMCS ReadFileHeader ifstream amp fin string keyword string info int t channels t bits t radix fin gt gt keyword if keyword channels return false fin gt gt info if Utility IsDecimal info t channels if t channels return false else return false fin gt gt keyword if keyword bits return false fin gt gt info if Utility IsDecimal info t_bits atoi info c str if t bits dE Sas StS return false 45 atoi info c str t channels t samplesPerSec r 16 bool XMCS Wri else return false fin gt gt keyword if keyword radix return false fin gt gt info if Utility IsDecimal info else fin gt gt keyword t radix atoi info c str if t radix t radix 16 return false return false if keyword format fin gt gt info if Utility Is
34. ist audioBuffer access audio blocks from audioBuffer and store to dataList SamplesPerSecond int tSamplesPerSecond Ay bool Play 7 bool Record Ef void AccessBlocks accessors x bool SetChannels int int GetChannels bool SetBits int int GetBits bool Set int Ge bool SetFileRadix int int GetFileRadix callback procedure for playback fel friend void CALLBACK waveOutProc HWAVEOUT hWaveOut UINT uMsg dwParam2 k endif DWORD dwInstance DWORD dwParaml DWORD 60 A 1 7 utility cpp include utility h char Utility Symbol int x switch x case 0 return 0 case 1 return qL case 2 return 2 case 3 return 3 case 4 return 4 case 5 return 5 case 6 return 6 case 7 return 7 case 8 return 8 case 9 return 9 case 10 return A case 11 return B case 12 return C case 13 return D case 14 return E case 15 return F default return 0 string Utility Int16ToHexString int num int bits if bits 8 amp amp bits 16 return char hex new char bits 4 1 int mask 0x00F0 0x000F 0xF000 0x0F00 int shift 4 0 12 8 for int i 0 i lt bits 4 i hex i Symbol num amp mask i gt gt shift i hex bits 4 0 string shex string hex delete hex return she
35. les back to their original seguence Taking for example an 8 point block of data with indices 0 7 using FFT using Decimation in Time Algorithm we can break down the whole block into the following TOT aio azi aa aie dur di dis al Note however that the indices of the new sequence that was produced when converted to their binary forms are just the bit reversed form of the original sequence Thus all calculations can be done in position if the original data sequence is shuffled already into the correct order the bit reversed form before passing it into the FFT algorithm 3 1 3 Fourier Class For this audio frequency analyzer project a Fourier class whose members and functions are enumerated below is created to provide functions that shall execute the FFT computation of our audio samples 3 1 3 1 Members The Fourier class includes the following members complex lt double gt data_array 1024 This is a member of the class that is used to store data samples so that 1t is readily processed by the functions inside the Fourier class int block length This is a member of the class that is used to store the block length of the data sequence 14 3 1 3 2 Functions The Fourier class includes the following functions void set int data int length This function sets or readies the data block it receives as a complex type array This function requires two parameters the data array which is t
36. mputation of the magnitude of the complex fourier transform double scale 1000 30000000 for int i 0 i lt block_length i mag i abs data array i scale 67 Bibliography Baterina Anna Veronica Macalintal Jeric xmcs player and recorderzip Internet Available from http groups yahoo com group admudsp files accessed July 2010 Decimation in Time FFT Algorithms Internet Available from http www ee caltech edu EE Courses EE32b handouts FFT pdf accessed July 2010 Fast Fourier Transform FFT Internet Available from http www cmlab csie ntu edu tw cml dsp training coding transform fft html accessed July 2010 Fast Fourier Transform Internet Available from http www mathcs org java programs FFT FFTInfo c12 2 pdf accessed July 2010 List Internet Available from http www cplusplus com reference stl list accessed August 2010 Koga Dan Extended wxDev C Tutorial 2009 Internet Available from http wxdsgn sourceforge net g node 11 accessed 2 Aug 2010 Programming with wxDev C 2006 Internet Available from http www physics ohio state edu ntg 780 downloads programming with wxDec C pdf accessed 13 Aug 2010 68 Smart Julian Roebling Robert Zeitlin Vadim Wx wxdc Internet Available from http docs wxwidgets org accessed August 2010 What are monophonic polyphonic and true tone ringtones 2009 Internet Available from http www cellutips com what are monoph
37. nd Mr Francis Tiausas for giving us advice and for helping us solve the problems that we encountered through the process Of course we extend our gratitude to our family especially our parents for their unconditional support and encouragement in our pursuits Without them we could not have gone to this point in our life in the first place All our regards and blessings we offer to all others who supported us in any respect during the completion of the project Table of Contents E IO UCM iaa duty sono Eee ass 6 Me Pee 6 1 2 Project Scope and imitations soon cones un Get 6 1 3 Overview of Components al See anke 6 2 XMCS SHR SOB ET use 8 2e AMOS Class anna bona AAA AK 8 2 1 L XMCSO COnStricton aia aiii 8 21 2 SMCS Record UA ida 9 2 1 3 XMCS Play TUCASA 9 2 1 4 XMCS Save and WriteBlockFromBuffer function 10 2 1 5 XMCS AccessBlocks TUN aiii iii 10 Ws INACTION CE NR 10 22 TEC AAS S cases tts ot doba TIT 12 3 Fourier CS A o da R bb 13 3 1 Theoretical Back Grounds edit lis 13 3 1 1 Decimation in Time Algorithm 13 3 12 BERNA eee 15 is Clas Scinn neonain a a a nt 15 SIMEN asr O O E ST 15 E A ee CORE 16 4 User Interface Implementation e ooo oo 17 4 1 qudiofrequencyanial yZer erin Mini isa 18 4 2 audiofrequencyanalyzerfrm cpp 19 42 1 Program FloW ueniens ige eai a Ris 19 SUser s Mardi i a po aaa Va ban ekon ada 21 5 1 CDROM C nten
38. ne note or one instrument at a time In contrast a polyphonic tone can play 40 notes at a time so it is expected to produce more freguency components than the monophonic tone Then the true tone is expected to display the most freguency components since it is a recording of actual music or sound cellutips com 2009 These expectations were more or less achieved and the proof are as follows Fig 6 5 1 Analysis of a monophonic ringing tone 32 Fig 6 5 2 Analysis of a polyphonic ringing tone Fig 6 5 3 Analysis of a true tone 33 6 6 Theoretical vs Actual Audio Blocks Processed To check whether the audio processing of the analyzer is real time or not we compared the actual number of audio blocks recorded using a timer and the indicator of audio blocks processed at the top of the GUI Note that the number of audio blocks that should be recorded if the recorder samples at 8000 samples per second with the audio block size set to 1024 samples is theoretically about 7 8 blocks So in 1 minute the number of blocks that should be recorded is 468 In our experiment the actual number of blocks processed in 59 67sec is 344 In 59 67sec the theoretical number of audio blocks is about 465 4 This means that about 121 blocks were missed by the program Conversely the program is able to process about 73 9 per minute And it only misses about 26 1 of the total theoretical audio blocks per minute The screen shot of the test is shown
39. onic polyphonic and true tone ringtones accessed 10 October 2010 69
40. re the length of the audio blocks is fixed to 1024 samples while the number of audio blocks to be processed is to be specified by the user 1 2 Project Scope and Limitations The recording part of the project supports only 16 bit XMCS audio and the sampling rate is fixed to 8 kHz The recording and displaying of its Discrete Fourier Transform is automatically looped where the number of loops is egual to the number of audio blocks specified by the user The recording could not be paused and then resumed It will also not stop until the recording has not reached the specified number of loops set by the user 1 3 Overview of Components In the development of the Audio Freguency Analyzer the following classes are utilized the XMCS class that provided the recording functionality that is needed to collect audio samples that are to be accessed for the DFT computation the Utility class that contained the functions needed to convert the 16 bit audio samples to a type that the DFT computation can handle the Fourier class that is used in executing the DFT computations and finally the audiofrequencyanalyzerfrm class for the development of the graphical user interface of the program 2 XMCS Player Recorder The parts used from the XMCS Player Recorder project from 2007 is discussed in this chapter The two classes used from this project are the XMCS class and the Utility Class In addition the main functions used are the record
41. rfrm class are predefined strings for the wx Widgets platform such as wxClientDC graph wxTextCtrl countAudioBlocks wxButton analyzeAudio wxStaticText numberOfAudioBlocks Similarly the constructors call on the functions inside the adjacent class 17 4 2 audiofreguencyanalyzerfrm cpp The class for the GUI is audiofrequencyanalyzerfrm audiofreguencyanalyzerfrm cpp is the location for most of the syntax of the program which calls all the functions from the necessary classes appended to it This code is mainly composed of two important function The output field is given by the wxEdit text field from the wxTextCtrl class and the application is enabled using wxButton1 Click 4 2 1 Program Flow The event is started when the user clicks the Analyze Audio button shown in Fig 4 2 2 Fig 4 2 2 Analyze Audio Button The number of blocks being processed is shown in the output field read in Number of Audio Blocks as shown in Fig 4 2 1 Number of Audio Blocks Fig 4 2 1 Input Field The default size of the audio block is set to 1024 samples The while inside the program is composed of several for loops This while loop is used to allow the program to process continuously The first function to be called inside the while loop is the record function by audiofreguencyanalyzerfrm audio Record The variables for the FFT function is declared next to the record function because some variables are ne
42. se wxCloseEventg event Destroy void audiofrequencyanalyzerfrm WxButton1Click wxCommandEvent event int Ctr while 1 record audio from soundcard audiofrequencyanalyzerfrm audio Record for FET int j 0 Fourier compute int data 1024 double out 1024 int np 1024 transfer audio blocks from linked list to array for int i 0 i lt np i data i 0 audio dataList clear audiofreguencyanalyzerfrm audio AccessBlocks for audio it audio dataList begin audio it audio dataList end audio it if j lt np data j audio it 38 JE take the FFT of arrays compute set data np compute FFT function compute magnitude out r graph gt Clear h graph gt DrawText Magnitude 5 50 graph gt DrawText Hz 1330 680 graph gt DrawText 0 3 660 graph gt DrawText 1000 325 660 graph gt DrawText 2000 660 660 graph gt DrawText 3000 995 a graph gt DrawText 4000 1315 660 graph gt DrawLine 5 650 2000 650 graph gt DrawLine 5 60 5 o int x 0 int y 0 for int i 0 i lt 513 itt x 5 1 2 6 y 650 out i 2 graph gt DrawLine x 650 x y graph gt DrawCircle x y 2 CEE TS countAudioBlocks gt Clear countAudioBlocks lt lt ctr 39 A 1 2 audiofreguencyanalyzerfrm h 1
43. t only which is the middle point F 8 For the FFT function it produced a result of 8 0 while the test code had 8 2 28666e 13 However since the difference is not that significant we can say that our FFT function is fairly accurate We also had a similar observation when the codes are tried out with 8 point samples With 0 to 7 as the input values the computation for the middle point F 4 in the original FFT function is 4 0 while for the test code we had 4 3 38746e 14 25 When the GUI was already created we tried comparing both codes again by displaying their graphs for 1024 input samples with values from 0 to 1023 It can be observed from the figures below that the graphs are more or less the same The two figures however are compressed versions of the original pictures captured the reason why the points are not clear Figure 6 2 4 1024 point freguency spectrum using the DFT test code 26 6 3 Constant Input Another test that we did is to generate a constant input and see if it produces an impulse function in the display Indeed a spike at point 0 was produced as can be seen from the figure below Figure 6 3 1 Impulse function in freguency spectrum 6 4 Sinusoidal Waves with Different Freguencies Using the software NCH Tone Generator we generated audible sinusoidal waves with different freguencies The freguencies generated were 500 Hz 1000Hz 1500Hz 2000Hz 27 2500Hz 3000Hz 3500Hz and
44. tS sic sto oa a aa SPS 21 NZI Wa S gt so o ud d ud k ro tun a p 6 22 5 3 Using the SEE 22 OTER EE 25 61 Lab VIEW strain altres 25 6 2 Decimation in Time FFT function vs DFT Test Code 23 6 3 Constitution 28 6 4 Sinusoidal Waves with Different Frequencies 28 6 5 Monophonic Polyphonic and True Tones eee 33 6 6 Theoretical vs Actual Audio Blocks Processed 35 01 Record TEA dos 35 7 Results and Recommend ations sr ee te ee dE 37 PSP STINK SAA AAA a WA tale dada hodu WEKA had 38 o A at kana do das 6 Kay py baza dan ake 38 Aisle SOUECE Placa ias 38 A 1 1 audiofrequencyanalyzerfrm cpp 38 A 1 2 audiofrequencyanalyzerfrm h 41 A 1 3 audiofrequencyanalyzerapp cpp 42 A 1 4 audiofrequencyanalyzerapp h 43 AD se ST 44 AGS vvs 59 ALTEA eee 62 AS OUEN A ekono dana 65 ALO fo rier hon WAA ara ao ada O a oo nv 66 BID MO SADA E A uo o P DS 69 1 Introduction 1 1 Project Objectives The main objective in this project is to create an Audio Freguency Analyzer which is an application that records and displays successively the freguency spectrum of audio samples The recording of the audio samples and the displaying of the freguency spectrum is intended to be done by blocks whe
45. ta waveHeader otherSet dwBytesRecorded LIST implementation audioBuffer insert audioBuffer end buffer do clean up for the other set before preparing for another recording if any wavelnStop hWaveIn otherSet waveInUnprepareHeader hWaveIn otherSet waveHeader otherSet sizeof WAVEHDR prepare a wave header for the other set to be used for the next recording if wavelnPrepareHeader hWaveIn otherSet waveHeader otherSet sizeof WAVEHDR MMSYSERR_NOERROR goto failed cleanup add buffer for the other set to be used for the next recording if wavelnAddBuffer hWaveln otherset waveHeader otherSet sizeof WAVEHDR MMSYSERR_NOERROR goto failed cleanup unless recording is stopped wait until recording buffer is full while waveHeader currentSet dwFlags 4 WHDR DONE if lastBufferSize 2048 56 getch waveInReset hWaveln 0 waveInReset hWaveln 1 isStopped true break take note of number of bytes actually recorded lastBufferSize lastByteCount waveHeader currentSet dwBytesRecorded cout lt lt lastBufferSize lt lt endl switch to the other set currentSet 1 currentSet currentSet 1 2 isFirst false while isStopped add last recorded buffer to the list buffer new char bufferSize otherSet 1 c
46. the following audiofreguencyanalyzerapp h audiofreguencyanalyzerapp rc and audiofreguencyanalyzerapp cpp which are used to implement the main program instructed from the audiofreguencyanalyzerfrm h and audiofreguencyanalyzerfrm cpp that are necessary to call the functions and execute them through the buttons placed on the audiofreguencyanalyzerfrm wxform The program is supported by the XMCS class and Fourier class The source code for these classes are completed with xmcs cpp xmcs h utility cpp utility h and fourier h 16 4 1 audiofreguencyanalyzerfrm h audiofreguencyanalyzerfrm h is the location for declaring global variables and constructors for the program The wxFrame platform is used in order to utilize the size of of the window Several wx classes are used in this program such as include lt wx dcclient h gt include lt complex gt All the drawing techniques for the GUI are contained inside the dcclient h Smart et al 2010 The complex class is included for the Fourier class The variable declared as public inside the audiofrequencyanalyzerfrm class are as follows XMCS audio void WxButtonlClick wxCommandEvent event The void function above is automatically inserted when the button is placed on the Frm wxform XMCS audio is a constructor for the XMCS class In this manner the variable audio is able to call the functions inside the XMCS class The variable declared as private inside the audiofreguencyanalyze
47. udioBuffer begin list lt char gt iterator ref buffer new char block size unless playback is stopped play everything contained in the list while bli audioBuffer end if kbhit getch waveOutReset hWaveOut break copy data to be used to buffer memcpy buffer bli block size ref bli if the last element of the list size may not be egual to the maximum if ref audioBuffer end block size lastBufferSize unprepare header to be used for playback in case not yet unprepared if waveBlocks waveCurrentBlock dwFlags amp WHDR PREPARED waveOutUnprepareHeader hWaveOut waveBlocks waveCurrentBlock sizeof WAVEHDR specify buffer length and audio samples waveBlocks waveCurrentBlock dwBufferLength sizeof char block size memcpy waveBlocks waveCurrentBlock lpData buffer sizeof char block_size prepare header to be used for playback if waveOutPrepareHeader hWaveOut amp waveBlocks waveCurrentBlock sizeof WAVEHDR MMSYSERR NOERROR return false play waveOutWrite hWaveOut waveBlocks waveCurrentBlock sizeof WAVEHDR 53 one less temporary buffer available EnterCriticalSection scriticalSection freeBlockCount LeaveCriticalSection criticalSection unless playback is stopped wait until there s a
48. urrentSet currentSet 1 2 memcpy buffer LPSTR waveHeader otherSet lpData waveHeader otherSet dwBytesRecorded audioBuffer insert audioBuffer end buffer CLEAN UP for int i 0 i lt 2 itt wavelnStop hWavelIn i waveInUnprepareHeader hWavelIn i amp waveHeader i sizeof WAVEHDR delete waveHeader i waveInClose hWaveln i lpData r return true failed cleanup for int i 0 i lt 2 itt wavelnStop hWaveln i waveInUnprepareHeader hWaveln i amp waveHeader i sizeof WAVEHDR delete waveHeader i waveInClose hWaveln i lpData return false 57 RR RR KK RK k kk k kk kk k kk k kk k kk AA JK CALLBACK Procedure for XMCS Audio Playback RR RR KR kok k k kk k kk kk k kk k kk k kk kk k AAA void CALLBACK waveOutProc HWAVEOUT hWaveOut UINT uMsg DWORD dwInstance DWORD dwParaml DWORD dwParam2 XMCS XMCS ptr XMCS dwInstance if uMsg WOM DON return FI EnterCriticalSection amp XMCS ptr gt criticalSection XMCS ptr gt freeBlockCount LeaveCriticalSection amp XMCS ptr gt criticalSection A 1 6 xmcs h ifndef XMCS H_ define XMCS H_ include utility h include lt windows h gt include lt mmsystem h gt include lt fstream gt include lt string gt include lt list gt include lt conio h gt using std string using std ifstream us
49. x string Utility Intl6ToBinaryString int num int bits if bits 8 amp amp bits 16 return 61 char binary new char bits 1 int mask 0x0080 0x0040 0x0020 0x0010 0x0008 0x0004 0x0002 0x0001 0x8000 0x4000 0x2000 0x1000 0x0800 0x0400 0x0200 0x0100 for int i 0 i lt bits i binary i Symbol num amp mask i mask i I 0 binary bits 0 string sbinary string binary delete binary return sbinary short int Utility HexStringToInt1l6 string hex if IsHex hex return 0 short int num 0 for int i llos hex length 1 i gt 0 i 2 for int j i 1 j lt i j int val switch toupper hex j case 0 val 0 break case 1 val 1 break case 2 val 2 break case 3 val 3 break case 4 val 4 break case 5 val 5 break case 6 val 6 break case 7 val 7 break case 8 val 8 break case 9 val 9 break case A val 10 break case B val 11 break case C val 12 break case D val 13 break case E val 14 break case F val 15 break default val 0 break if unknown digit assign 0 num num 16 val return num 62 short int Utility BinaryStringToIntl16 string bin if IsBinary bin return 0 short int num 0 int mark bin length gt 8 bin length
Download Pdf Manuals
Related Search