MATLAB mfcc gmdistribution fit for Speech Recognition Program

Question

I'm new to Matlab and doing a signal processing project(Speech Recognition). After doing some calculations, I get some values known as MFCC (Mel-Frequency Cepstral Coefficient) in a matrix. I'm now supposed to apply a Gaussian Mixture Model (GMM) distribution using the function gmdistribution.fit(X,k). But I keep getting the error,

X must have more rows than columns.

I don't understand, how can I fix this? I tried doing a transpose of the matrix, but then, I get other errors.

??? Error using ==> gmcluster at 181
Ill-conditioned covariance created at iteration 3.

Error in ==> gmdistribution.fit at 199
    [S,NlogL,optimInfo] =...

My MFCC matrix generally has 13 rows and about 50-80 columns.

Any ideas on how to fix this? Should I be using only upto 12 cols at a time? OR what could be an alternate expectation-maximization (EM) algorithm to obtain a maximum likelihood (ML)estimate in Speech recognition?

Here's a sample matrix that I get after extracting the mfcc feature vectors from the speech:

 53.19162380493035  53.04536473593154   52.52404588266867   52.76558091790412   53.63907256262721   53.357790132994836  52.73205096524416   52.902995065027056  52.61096061282659   54.15474467851871   53.67444472478125   52.64177726437717   52.51697384592561   52.71137919365186   53.092851922453896  53.16427640450918   54.43019514688636   60.79640902129941   59.84919922646779   63.15389910551327   61.88723594060794   64.74826830389657   64.8349874832628    64.86278444375218   65.76126193531795   65.64589407152897   65.46920375829764   65.69178734432299   65.28831375816117   64.56074008418904   63.4966945660873    63.81859800557705   63.72800219675504   62.48994205815299   62.170438508902436  61.06563184036766   59.13583014975035   58.81335869501639   56.32130498897641   55.13711899166046   54.013505531107796  54.15759852717166   53.44176740036524   53.13219768600348   53.03407270007307   52.88271825256845   53.822163186509016  53.53892778841879   54.04538463287215   59.485371756367954  58.48009762761471   54.643413468895346  52.808848460884654  52.87392859698496   52.42111841679119   53.2365666558251    53.30622484832905   53.1799318016215    53.784807994410315  53.248067707554924  52.69122098296521   52.50131276155125   53.43030515391315   53.902384536061604  54.029570128176985  52.842675820980034  52.79731975873874   53.18695701339912
-10.209801833131205 -9.680631918902254  -9.62767876068187   -11.100788671331799 -12.214764051532008 -10.968305830999338 -9.860973825750351  -9.865056435511548  -10.658715794299441 -9.3596215435813    -11.6646716335442   -11.73183849207276  -12.378134406457027 -10.926012890327158 -11.620321504456165 -10.158285684702548 -9.264017760124812  -3.477686356268614  -3.34008367962826   -4.830538727398767  -2.000396004172366  -4.4851181728969225 -2.9033880784025152 -4.367902167404347  -4.497084603581041  -5.199683464056032  -5.906443970301479  -6.1194300184632855 -5.96250940992931   -6.359811770556116  -6.264817939973589  -4.895405335125048  -5.356838360441918  -6.327382452484718  -6.680325151391659  -6.17848037726304   -5.4759013940523245 -1.9841026636312946 -4.076294540940979  -7.824603409725002  -5.800269620602235  -8.01263214623702   -11.425250071230579 -10.277472714265365 -10.774573945280718 -11.322162485376891 -10.052477908307408 -10.004482396755566 -8.557096237262265  -7.319189335399103  -4.798868632345757  -10.203105092807693 -10.406716632774856 -11.067414745093817 -11.699111553041329 -10.749597806292954 -10.555273429092225 -8.854304279940754  -10.903698849240602 -10.234951031082241 -11.550994106255267 -11.295232804215324 -10.688554946454785 -9.208980407123816  -10.585845595336993 -10.757300448605834 -10.319608162526984 -10.551598424355781
-0.18311276580153307    -1.3000235617058096 -2.379404485976171  0.8537711039288245  0.7835891293988151  -0.786100291329253  1.0107138900981782  -0.12469382941718324    -2.2952791566222173 -0.8251663787748776 -0.050658777310996696   -4.6807361290865295 -3.3756455575107784 0.38895610612101605 -0.9962664893365839 -1.3680101462804826 -0.7328675082528926 14.930618844131613  11.172961105935304  16.974801313922335  13.375385369069916  14.024700863057664  14.594849346714536  17.610029847404075  16.601731375214815  15.581203919095396  15.429198596491359  15.842389728372694  16.162847697063377  17.262648834400064  18.2608582394078    19.38844125300681   16.858591012785013  16.93154670795065   12.906259456599424  13.056739996060314  11.258250889980491  8.834726263239137   6.184939770895715   4.068236554570518   2.184520358080839   3.6716311416454106  0.5890504959921528  -3.0455374126328874 -1.657407892408495  0.33660057466143056 -0.40801030148804557    0.04270808730635576 3.208411924734062   5.821481390407001   4.560967865706884   -0.9575473658761547 -1.9690622742411314 -1.4335363449433605 0.5073073427521086  1.8313651620152203  -2.1659200593772345 1.2769675752335854  -2.2873258303700696 -0.030049578085935582   -2.002440722711317  -2.3424337647822346 -4.259810095095228  -0.9747655920995262 0.09482704525635513 -0.2885341356828254 1.439149953470075   0.6807611595304401
2.087244713218005   -3.787403802296573  -4.665688240227797  0.46022874550890147 -0.16943798737784035    -2.7170563621342785 -1.7464303367036695 -3.27442943105816   -3.6318990907200597 -1.1574346481702122 -1.0207450052082863 -5.838249114276465  -4.864029691290982  -2.7443279494466704 -1.3475670289669839 -0.71926223394222   -1.7145131082739746 10.695036462762722  10.398176627688748  11.642258160333318  8.67660434911699    13.223576542483247  14.470121526018994  14.100543157086074  13.22291384069529   11.67823582796623   13.466476916853203  13.535357097626715  14.875339057135838  14.37083096189283   13.33673313953938   12.329553090328996  9.676373050790103   11.448653427990415  9.874926564656558   7.147530590070999   10.29584390330658   10.101141207939456  5.283325337013565   4.507665609590605   3.1555597807254223  1.176891149051998   -0.2017066100725112 -2.5074705794245427 3.7132131484813073  0.9607407688505634  -3.2742739297063865 -6.602070936837743  -2.2912280318564378 10.190482148210974  10.157945177713376  -0.09147003586407224    -5.244432802624313  -1.2872483780850776 -3.7378553488851147 2.853534940706138   -2.9599246290596257 -1.2759697907404983 -2.609173347676013  -0.027021884588768103   -2.3092682012995387 -1.4002697262020989 -4.192442987678205  -0.11708538059933485    -1.722764980370641  -0.8528543327485958 0.36818682029243044 -1.5833959315094956
-1.2340033668089612 -2.7554310519289933 1.4704457874837413  -0.4125243211298726 1.7297567688324673  3.721374587353874   -2.2232745236466402 -1.0295891117338212 1.021098021933131   -3.392544522126444  1.3301447592375433  -0.30182589581098784    -2.2645887723031413 0.5179073904608001  2.0537130718040917  -3.030349632233867  -2.107849434880047  -7.949976055283274  -5.172658838436902  -7.2904509401269575 -6.1323858833603815 -2.37546696444418   -2.6620539778383723 -3.5795807500300305 -4.687709564035536  -1.7454933814935076 -0.6827757483935794 0.23687223893178067 2.8267871613253077  3.5866135581831227  3.142665641927276   4.095262325494299   3.871285159350548   3.8703187080829764  3.8314236250858555  1.798983626211966   0.725468180389042   0.11919814479647405 2.7173707003940124  6.868690477210499   6.270964718280218   2.3176609494750564  2.0733820130334926  -0.8539453920978304 3.48931978155834    -2.6098957232427957 0.7925129692289851  -2.482250690121881  -1.9255950956807195 -3.3296568338000525 -2.5852039200206076 0.7513494304110043  1.6119079892129162  0.8581457406304087  1.4037071284373093  -3.163651849398714  5.052978402873416   2.4518824480379813  0.027602305580521395    0.7477958990121767  0.9232542431737198  -0.5545479544994354 -3.4480660326803503 1.0747263160741485  -4.078097840161742  4.485742151839941   0.1658605159666291  0.1722930547996016
-1.6428664752690114 3.7865726986742145  2.5318491820052564  -2.1947219298888676 -2.1237775233625986 2.598630953202959   -6.076201524281277  -5.315246911864284  -1.5747455209374586 -3.223379488606859  2.6008295264581776  1.3270506534986315  -2.5790744715346676 0.7756431623687378  3.0553271757777356  -0.20800002044634847    -1.530027153710214  -2.207970121996219  -1.8813636939941347 -2.685201388968379  -1.2497372042225408 2.5726591149003712  1.4779209530617206  0.18848939011950389 -0.8737068656038859 4.364271583896629   2.0338276700410187  4.017665258617117   2.929288856255161   10.031463178073729  7.807148474194119   8.930649791195147   9.356704480964387   4.682860624638529   3.9421955431659375  3.46979114616638    0.10907941624689588 1.013539556043216   1.380950812959332   1.077296756517698   4.643176114193134   0.276532579753215   1.3247848485761091  -1.6452351331258643 5.459080479943587   -2.623903958160855  -3.6495250981385525 0.30098983943901886 1.2192582165344557  3.9341748890807207  3.8902438441040768  2.3070835920696586  -2.692501110699399  1.6807838025217028  1.5259881694196216  0.3750392433389195  5.708674336592535   -1.1571072509634228 -1.9909829706185518 -2.911287549300028  -4.934348834333174  -2.258176779559039  0.17624511060134188 0.02295826196619305 -3.516972940169973  5.184345513656031   1.4594074325337887  -0.19794455729474633
2.362306464828889   1.8140886321872307  3.105122487428386   -2.452729932993756  -1.9482153346221507 0.23556664481369372 1.0605939999557794  9.466891504042334   4.485454438679325   2.6792667132201102  -0.7696085536288818 1.1799363148487811  -4.770207147524265  0.7773255533610134  -1.0253054017942649 5.364238239319841   3.1331011184169473  4.744685304867839   -0.052537238369118014   4.477806263589113   3.1539530991186067  6.4185233259645385  2.549990446321861   2.4829837421356564  4.089323590949597   7.9396405004582045  6.041498345508568   9.234608707932582   7.3843205505399885  10.495371462065135  15.043508733932194  8.70736248600434    13.199534350054295  9.807690741908354   9.182134815924455   12.06839623216329   7.974743468866006   12.349726591545481  5.750367027892127   -0.6482940009399485 5.4638120941442185  1.856389413910232   1.9530813300592067  -2.8701346921179733 1.558852931425583   -0.19366384484174437    -2.6386457918474457 1.4662219452543457  2.079641671534525   15.326629935694294  14.705559998054612  -0.06282946858494885    -1.827803410621235  3.114649202395378   0.3720781976421628  0.43011998686353536 -3.376799358785071  -1.5552531679484054 3.060902156478365   3.5360394473034553  -2.3908283396567356 0.6675611086499327  0.22711502816964574 -6.457828495248154  -0.6807474446526474 -0.6230980701736715 2.2692316872172476  -0.979235567032777
2.306823535295793   3.4952484194762055  5.910905884417197   -3.0627994884681873 -3.2217585242174294 -0.015187803494101149   -0.9514287527346498 3.114431724585367   0.42923281798814705 -3.189859804015462  -1.472673603923648  -3.036867739556342  -0.15973786580917693    -0.0905525722541792 2.330382174351248   2.7439958525955515  0.3730263667251821  -12.515523622378907 -13.343548342714616 -11.536760383050373 -8.307383651556634  -15.660481772806875 -14.155076207607415 -14.343032997039627 -11.791205489191787 -14.964231411185601 -13.183950294156357 -8.972526839374074  -5.366478645304655  -10.910217774510665 -1.5480767893424763 -8.888577773693916  -2.6255911360834023 -5.8588628908556695 -4.145564000313309  -2.984375697431632  0.8831077064431804  -5.243824833303439  5.196626588048474   6.352837095147023   1.2112116324076188  -2.9147691775934286 -2.6935780565318352 -2.810972986669758  4.9399646272914275  -1.1703117105056318 -2.402532372315127  -4.8461309660884675 -7.261524451953783  -2.5282219889051856 -1.0065282601086587 -2.5563997598612156 -4.351683980269447  -0.46252498899381495    -5.890633052969005  -0.3032076532083649 -0.6457938679695084 -0.455043482005029  3.359840875612215   -1.7228176367513395 -3.168976094613273  -2.5233843488620917 -6.495499983402964  -3.4972987525688515 0.7115283186290751  -2.581097605905542  0.6315410714331887  0.19502062594451325
1.2870172739850947  2.713157481924801   -0.5205380954882455 -4.658525381198428  -0.10827507866220412    2.4486415136057875  -0.2640204926534809 -0.09970608992954652    1.5082258768440102  -0.48148890836461583    6.911722876338505   -1.839425896561688  -3.841669694063511  -4.524554996776859  -0.9323811218879002 -6.12813923896959   -2.617633134059251  -6.309717724130619  -3.909047191185573  -6.705305972326263  -3.194505292603528  -7.893721876340621  -0.7610949447938617 -0.6090909340423546 1.4581855733113227  -2.41596099072141   -3.8541389118806912 -1.927700181895679  4.665459793274741   -2.132645903487048  4.157947245063189   0.11326683589817262 -1.162075689787945  1.055761599597126   3.298475882289032   0.9391848013866494  5.223274229835592   5.199193224601442   6.24812913948699    5.2190463423872515  1.5179114498579496  -0.6790185492512775 -0.31373376397636593    -3.5993965276962707 4.302535367682559   5.0068035330847005  -2.436072054028143  -0.8350201387276532 -2.018104375721472  0.5404586080558861  -2.428770201558009  -2.335732881592787  -0.052034561490399235   2.6353099398265676  -2.99995676341149   1.7399565653589897  -0.29483744276382473    4.957413374961816   5.6898464888615 -4.002464222625706  0.966133847419872   2.170532357744949   -2.4172124815273173 -5.913083394982123  -0.22652498917043715    0.138040634076645   -2.826152803587723  5.842509989192995
-6.149578124267104  -2.8288721761218962 -0.27284674336933024    -0.7388702321118317 6.111878602550777   3.359125556152289   1.2074835809541602  -2.229103203811113  -1.625118718284933  -0.2004222132512952 2.0932748099429754  0.712406626137792   0.43416711590137985 -5.55554193439384   -2.1786650973628827 -2.969057723871395  -6.618199451327406  2.299416281672153   5.007013248892597   -2.8033104103688027 -0.14925301159195922    -3.1533724522208697 -1.686316186073986  -0.08884837954280254    1.3265208802169017  1.3523930289041641  3.5524134648371395  -1.4254466520590146 -3.5611240333626477 1.0329276937146186  0.753052597154297   0.7975894394949765  -1.1854014340942607 1.1593797963914545  -0.8529267167794818 -5.171015036219429  4.116322136411159   -1.4483994704782983 -4.286164521201809  2.740046108799948   -3.5798763236060673 -3.018292657641495  -3.1806602684198966 -7.234273046469597  -9.434807181114692  -1.5847563989433828 -3.5635243742856346 4.782665786942992   3.5778211425622497  1.6853638633605281  1.7167799803768633  -1.6174055012561088 3.7435401900571574  -3.176593678259591  6.40495736593622    3.3331406463423483  -4.189245091250336  -1.1362166265192732 4.592859698246665   -2.8863334811724606 0.16041676401714375 4.737837256397985   -2.2744510630052366 1.4695485402180768  -4.897075450622638  1.0194864096015128  3.0757846367935398  1.489203230013674
-5.616870225243653  4.10940999519677    -0.3567822711722583 4.987855490462697   2.5632059692246143  -4.705396196410884  -0.1194996962733683 8.46869233605413    1.7788275688487483  -1.9527299063266377 -1.481085011956697  -1.0244613136295895 3.2992905241167114  -3.64385218716246   0.4426619512128128  -0.9239334997116153 -1.8620760850713798 -1.572039531941818  -10.036763755809012 -4.991528131941471  -7.136095340914314  -3.9318863449619683 -8.239368103131268  -8.443697887490892  -7.638579800501108  -8.460278636486919  2.042450826339361   -2.9885807367329646 -7.09364471308204   0.751496922690038   -0.7845673603407124 3.01935526513198    -1.39022538332522   -1.3101410638362037 -6.557786354682332  -10.172228179790066 -7.914321004354581  -5.649458806929109  2.0908760762554857  -1.4736963383710477 -1.1834278800206155 -0.6892124083994282 4.710875739605662   -3.269448539379895  -1.365967094144594  2.229881555767406   -0.9419137895352326 -0.48671864439322476    4.178896930726449   -6.953289505262448  -3.5225552311666406 -0.03841148260907753    0.14013269702442782 -6.512368259808616  1.8280649782849192  0.3454330974085145  -7.766620058704248  1.6650823954773208  9.615187994533223   3.360235349725343   0.22182808924480077 -0.30209172650913635    -1.6349262462057823 5.754809401078592   -1.6377375938940244 4.58705098784457    -2.404590707062002  0.45319882935997813
-3.730821551088958  1.1493694300690667  6.12342052964259    1.0160737493461047  4.543231805847945   -0.46099872305259204    -1.5594323941163388 10.090773095751917  5.028250117132579   1.5903687490782517  0.5749808655709501  -4.492674335179201  2.325703447395548   5.206408565021089   -4.9872461967223565 -6.549149325309605  2.90139977554803    -3.116490551862926  -8.703818668102071  -0.6313375630613844 -1.3155034176934333 1.1556044127857454  -0.9275062964334158 2.1324193244502876  3.430145051864411   4.086699745467884   5.480203425684989   1.3741912885959398  3.339835767680544   5.640295156144797   1.9610369474663063  1.785080274117643   1.8291947445479142  2.966205980470809   -0.12596430958161875    4.646073914100102   -0.7648039700071241 6.3484330647888605  4.459704396949977   1.5062484187054803  -1.6168718590653306 1.7558262745105164  1.2355091938620948  -9.312287204368275E-4   -0.5174901532050828 -3.0942917590395123 2.127834965233185   2.205667503405521   1.120114080459297   -1.7595270682165296 -9.083346980110788  -1.4981626322158839 0.7146008123272161  -0.6811098332417078 0.32703395934824275 -2.555380698176684  1.7740823756697832  4.5707670000209495  1.4842964294571344  4.818614788487457   3.1215801329358515  1.4479667080737233  1.1758507462380035  6.03230783411774    2.288914057777  4.82860171466599    -1.2457175363287405 -0.5058301430711261
-2.768473705667538  0.15564719507110275 -2.6550122323991947 -5.709488621527887  0.4785386384778287  0.6814858260993006  -5.52429514744985   -0.5602195429716864 3.9723119003523184  -5.62516538263036   -4.829570651115459  -1.2950948013109767 7.302412416568166   -3.043678812305364  -3.149850274277347  -6.476944546181209  -0.5807442791158823 -4.080078654055604  -3.1611933621382597 -0.11637063086775598    1.6049131611665592  5.044497534034215   0.3838925988521055  5.778293566481567   4.058620434329893   5.927479580737815   2.489198330275847   1.3107947997423626  1.5828295303331719  0.024839158566965516    -0.5476121359730696 0.87259267290178    0.9361180475548712  -1.5960762918622518 -5.611058251792273  -0.1594321010434905 -4.760816879788385  -0.07479939429503339    -1.7483043512234622 -2.8457787380793556 -1.7121754676101464 -3.787278050262899  3.7473965097918542  -1.659644247031472  -0.09111384850703107    2.4558095815874137  -0.06434581404575994    -3.7711115877495898 -0.2647997786903864 7.047915131872554   2.696723847584077   2.0890029827477234  -1.6825745638184928 -3.5887592066629557 -1.6594244317183802 -3.1951431164448874 3.27560938604933    2.334479543234365   2.9783519550285447  4.899933974871159   -2.2328606908007633 1.600105125583785   -2.1591853807024437 5.713548445622229   2.1891014794399264  -4.680943918675132  -2.5283217348396123 -2.6580555791689666

Answer 1

I don't understand, how can I fix this?

You need to transpose the matrix, you got it right. The vectors must be on the raws

Any ideas on how to fix this?

GMDISTRIBUTION implements the standard Expectation-Maximization (EM) algorithm. In some cases, it may converge to a solution which contains singular or close-singular covariance matrix for one or more components. Those components usually contains a few data points almost lying in a lower-dimensional subspace. A solution with singular covariance matrix is usually considered as spurious. Sometimes, this problem may go away if you try another set of initial values; Sometimes, this problem will always occur because of any of the following reasons:

• The number of dimension of data is relatively high, but there are not enough observations.
• Some of the features(variables) of your data are highly correlated.
• Some or all the features are discrete.
• You try to fit the data to too many components.

In your case, it seems that the number of components that you used, 8, is too big. you can try to reduce the number of components. Generally, there are also other ways that you can use to avoid getting "Ill-conditioned covariance matrix" error message"

1. If you don't mind to get solutions with ill-conditioned covariance matrix, you can use option 'Regularize' in the GMDISTRIBUTION/FIT function to add a very small positive number to the diagonal of every covariance matrix.
2. You can specify the value of 'SharedCov' to be true to use equal covariance matrix for every component.
3. You can specify the value of 'CovType' to be 'diagonal' .

See also

http://www.mathworks.com/matlabcentral/newsreader/view_thread/168289

Should I be using only upto 12 cols at a time?

No

Answer 2

@Shark I had the same problem, trying to generate an Gaussian MM object from a set of data. I solved it by specifing the type of covariance:

GMM1=gmdistribution.fit(X,k,'CovType','diagonal')

GMM1 is the object name. You can find the meaning of X and k in

help gmdistribution.fit

If this doesn't work for you, try specifying the initial values of the EM algorithm that gmdistribution already uses to generate the GMM.

Elios

Answer 3

first of you should make it linear because it is too big for matlab to do it, after that its better to just take 7-10 features (i think you get more than this). after you did your work then use reshape function in order to make it what you want