Pocketsphinx переполнен, вызовы чтения слишком редки - PullRequest
1 голос

Я пытаюсь установить Pocketsphinx на динамик, который работает с Buildroot.Перед тестированием на динамике я попробовал Pocketsphinx на своем Raspberry-pi3.Проблема в том, что у меня разные результаты на обоих устройствах, несмотря на то, что они имеют примерно одинаковую производительность оборудования.

Когда я пытаюсь декодировать файл WAV, у меня есть этот результат на обоих устройствах (и точность, и время одинаковы):

pocketsphinx_continuous -infile /mnt/usb/file-16000.wav -dict /mnt/usb/sphinx-
french/fr.dict -lm /mnt/usb/sphinx-french/fr-small.lm.bin -hmm /mnt/usb/sphinx-f
rench/cmusphinx-fr-5.2
INFO: pocketsphinx.c(152): Parsed model-specific feature parameters from /mnt/usb/sphinx-french/cmusphinx-fr-5.2/feat.params
Current configuration:
[NAME]                  [DEFLT]         [VALUE]
-agc                    none            none
-agcthresh              2.0             2.000000e+00
-allphone
-allphone_ci            no              no
-alpha                  0.97            9.700000e-01
-ascale                 20.0            2.000000e+01
-aw                     1               1
-backtrace              no              no
-beam                   1e-48           1.000000e-48
-bestpath               yes             yes
-bestpathlw             9.5             9.500000e+00
-ceplen                 13              13
-cmn                    current         current
-cmninit                8.0             8.0
-compallsen             no              no
-debug                                  0
-dict                                   /mnt/usb/sphinx-french/fr.dict
-dictcase               no              no
-dither                 no              no
-doublebw               no              no
-ds                     1               1
-fdict
-feat                   1s_c_d_dd       1s_c_d_dd
-featparams
-fillprob               1e-8            1.000000e-08
-frate                  100             100
-fsg
-fsgusealtpron          yes             yes
-fsgusefiller           yes             yes
-fwdflat                yes             yes
-fwdflatbeam            1e-64           1.000000e-64
-fwdflatefwid           4               4
-fwdflatlw              8.5             8.500000e+00
-fwdflatsfwin           25              25
-fwdflatwbeam           7e-29           7.000000e-29
-fwdtree                yes             yes
-hmm                                    /mnt/usb/sphinx-french/cmusphinx-fr-5.2
-input_endian           little          little
-jsgf
-keyphrase
-kws
-kws_delay              10              10
-kws_plp                1e-1            1.000000e-01
-kws_threshold          1               1.000000e+00
-latsize                5000            5000
-lda
-ldadim                 0               0
-lifter                 0               22
-lm                                     /mnt/usb/sphinx-french/fr-small.lm.bin
-lmctl
-lmname
-logbase                1.0001          1.000100e+00
-logfn
-logspec                no              no
-lowerf                 133.33334       1.300000e+02
-lpbeam                 1e-40           1.000000e-40
-lponlybeam             7e-29           7.000000e-29
-lw                     6.5             6.500000e+00
-maxhmmpf               30000           30000
-maxwpf                 -1              -1
-mdef
-mean
-mfclogdir
-min_endfr              0               0
-mixw
-mixwfloor              0.0000001       1.000000e-07
-mllr
-mmap                   yes             yes
-ncep                   13              13
-nfft                   512             512
-nfilt                  40              25
-nwpen                  1.0             1.000000e+00
-pbeam                  1e-48           1.000000e-48
-pip                    1.0             1.000000e+00
-pl_beam                1e-10           1.000000e-10
-pl_pbeam               1e-10           1.000000e-10
-pl_pip                 1.0             1.000000e+00
-pl_weight              3.0             3.000000e+00
-pl_window              5               5
-rawlogdir
-remove_dc              no              no
-remove_noise           yes             yes
-remove_silence         yes             yes
-round_filters          yes             yes
-samprate               16000           1.600000e+04
-seed                   -1              -1
-sendump
-senlogdir
-senmgau
-silprob                0.005           5.000000e-03
-smoothspec             no              no
-svspec
-tmat
-tmatfloor              0.0001          1.000000e-04
-topn                   4               4
-topn_beam              0               0
-toprule
-transform              legacy          dct
-unit_area              yes             yes
-upperf                 6855.4976       6.800000e+03
-uw                     1.0             1.000000e+00
-vad_postspeech         50              50
-vad_prespeech          20              20
-vad_startspeech        10              10
-vad_threshold          2.0             2.000000e+00
-var
-varfloor               0.0001          1.000000e-04
-varnorm                no              no
-verbose                no              no
-warp_params
-warp_type              inverse_linear  inverse_linear
-wbeam                  7e-29           7.000000e-29
-wip                    0.65            6.500000e-01
-wlen                   0.025625        2.562500e-02

INFO: feat.c(715): Initializing feature stream to type: '1s_c_d_dd', ceplen=13, CMN='current', VARNORM='no', AGC='none'
INFO: cmn.c(143): mean[0]= 12.00, mean[1..12]= 0.0
INFO: acmod.c(152): Reading linear feature transformation from /mnt/usb/sphinx-french/cmusphinx-fr-5.2/feature_transform
INFO: mdef.c(518): Reading model definition: /mnt/usb/sphinx-french/cmusphinx-fr-5.2/mdef
INFO: bin_mdef.c(181): Allocating 101051 * 8 bytes (789 KiB) for CD tree
INFO: tmat.c(149): Reading HMM transition probability matrices: /mnt/usb/sphinx-french/cmusphinx-fr-5.2/transition_matrices
INFO: acmod.c(113): Attempting to use PTM computation module
INFO: ms_gauden.c(127): Reading mixture gaussian parameter: /mnt/usb/sphinx-french/cmusphinx-fr-5.2/means
INFO: ms_gauden.c(242): 2108 codebook, 1 feature, size: 
INFO: ms_gauden.c(244):  8x32
INFO: ms_gauden.c(127): Reading mixture gaussian parameter: /mnt/usb/sphinx-french/cmusphinx-fr-5.2/variances
INFO: ms_gauden.c(242): 2108 codebook, 1 feature, size: 
INFO: ms_gauden.c(244):  8x32
INFO: ms_gauden.c(304): 0 variance values floored
INFO: ptm_mgau.c(804): Number of codebooks exceeds 256: 2108
INFO: acmod.c(115): Attempting to use semi-continuous computation module
INFO: ms_gauden.c(127): Reading mixture gaussian parameter: /mnt/usb/sphinx-french/cmusphinx-fr-5.2/means
INFO: ms_gauden.c(242): 2108 codebook, 1 feature, size: 
INFO: ms_gauden.c(244):  8x32
INFO: ms_gauden.c(127): Reading mixture gaussian parameter: /mnt/usb/sphinx-french/cmusphinx-fr-5.2/variances
INFO: ms_gauden.c(242): 2108 codebook, 1 feature, size: 
INFO: ms_gauden.c(244):  8x32
INFO: ms_gauden.c(304): 0 variance values floored
INFO: acmod.c(117): Falling back to general multi-stream GMM computation
INFO: ms_gauden.c(127): Reading mixture gaussian parameter: /mnt/usb/sphinx-french/cmusphinx-fr-5.2/means
INFO: ms_gauden.c(242): 2108 codebook, 1 feature, size: 
INFO: ms_gauden.c(244):  8x32
INFO: ms_gauden.c(127): Reading mixture gaussian parameter: /mnt/usb/sphinx-french/cmusphinx-fr-5.2/variances
INFO: ms_gauden.c(242): 2108 codebook, 1 feature, size: 
INFO: ms_gauden.c(244):  8x32
INFO: ms_gauden.c(304): 0 variance values floored
INFO: ms_senone.c(149): Reading senone mixture weights: /mnt/usb/sphinx-french/cmusphinx-fr-5.2/mixture_weights
INFO: ms_senone.c(200): Truncating senone logs3(pdf) values by 10 bits
INFO: ms_senone.c(207): Not transposing mixture weights in memory
INFO: ms_senone.c(268): Read mixture weights for 2108 senones: 1 features x 8 codewords
INFO: ms_senone.c(320): Mapping senones to individual codebooks
INFO: ms_mgau.c(144): The value of topn: 4
INFO: phone_loop_search.c(114): State beam -225 Phone exit beam -225 Insertion penalty 0
INFO: dict.c(320): Allocating 109102 * 20 bytes (2130 KiB) for word entries
INFO: dict.c(333): Reading main dictionary: /mnt/usb/sphinx-french/fr.dict
INFO: dict.c(213): Dictionary size 105003, allocated 1018 KiB for strings, 1375 KiB for phones
INFO: dict.c(336): 105003 words read
INFO: dict.c(358): Reading filler dictionary: /mnt/usb/sphinx-french/cmusphinx-fr-5.2/noisedict
INFO: dict.c(213): Dictionary size 105006, allocated 0 KiB for strings, 0 KiB for phones
INFO: dict.c(361): 3 words read
INFO: dict2pid.c(396): Building PID tables for dictionary
INFO: dict2pid.c(406): Allocating 36^3 * 2 bytes (91 KiB) for word-initial triphones
INFO: dict2pid.c(132): Allocated 15696 bytes (15 KiB) for word-final triphones
INFO: dict2pid.c(196): Allocated 15696 bytes (15 KiB) for single-phone word triphones
INFO: ngram_model_trie.c(347): Trying to read LM in trie binary format
INFO: ngram_search_fwdtree.c(74): Initializing search tree
INFO: ngram_search_fwdtree.c(101): 742 unique initial diphones
INFO: ngram_search_fwdtree.c(186): Creating search channels
INFO: ngram_search_fwdtree.c(323): Max nonroot chan increased to 122777
INFO: ngram_search_fwdtree.c(333): Created 733 root, 122649 non-root channels, 124 single-phone words
INFO: ngram_search_fwdflat.c(157): fwdflat: min_ef_width = 4, max_sf_win = 25
INFO: continuous.c(307): pocketsphinx_continuous COMPILED ON: May 27 2019, AT: 10:26:52

INFO: ngram_search.c(459): Resized backpointer table to 10000 entries
INFO: ngram_search.c(467): Resized score stack to 200000 entries
INFO: ngram_search.c(459): Resized backpointer table to 20000 entries
INFO: cmn_prior.c(131): cmn_prior_update: from <  8.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00 >
INFO: cmn_prior.c(149): cmn_prior_update: to   < 32.09  6.52  5.60  4.29 -3.16 -8.27 -12.06 10.93  2.41 -0.40 -8.79  0.44 -0.56 >
INFO: ngram_search_fwdtree.c(1550):    14013 words recognized (74/fr)
INFO: ngram_search_fwdtree.c(1552):   215554 senones evaluated (1134/fr)
INFO: ngram_search_fwdtree.c(1556):  1691781 channels searched (8904/fr), 88058 1st, 244061 last
INFO: ngram_search_fwdtree.c(1559):    25969 words for which last channels evaluated (136/fr)
INFO: ngram_search_fwdtree.c(1561):   232530 candidate words for entering last phone (1223/fr)
INFO: ngram_search_fwdtree.c(1564): fwdtree 7.54 CPU 3.968 xRT
INFO: ngram_search_fwdtree.c(1567): fwdtree 7.54 wall 3.971 xRT
INFO: ngram_search_fwdflat.c(302): Utterance vocabulary contains 616 words
2019-05-28 12:25:39.086 [  3] E MessageRouter:sendFailed:reason=noActiveTransport
INFO: ngram_search_fwdflat.c(948):    10755 words recognized (57/fr)
INFO: ngram_search_fwdflat.c(950):   117062 senones evaluated (616/fr)
INFO: ngram_search_fwdflat.c(952):   610492 channels searched (3213/fr)
INFO: ngram_search_fwdflat.c(954):    60608 words searched (318/fr)
INFO: ngram_search_fwdflat.c(957):    31719 word transitions (166/fr)
INFO: ngram_search_fwdflat.c(960): fwdflat 2.83 CPU 1.491 xRT
INFO: ngram_search_fwdflat.c(963): fwdflat 2.84 wall 1.492 xRT
INFO: ngram_search.c(1197): </s> not found in last frame, using lin.188 instead
INFO: ngram_search.c(1250): lattice start node <s>.0 end node lin(2).106
INFO: ngram_search.c(1276): Eliminated 61 nodes before end node
INFO: ngram_search.c(1381): Lattice has 851 nodes, 10478 links
INFO: ps_lattice.c(1380): Bestpath score: -5096
INFO: ps_lattice.c(1384): Normalizer P(O) = alpha(lin(2):106:188) = -441437
INFO: ps_lattice.c(1441): Joint P(O,S) = -466741 P(S|O) = -25304
INFO: ngram_search.c(872): bestpath 0.34 CPU 0.182 xRT
INFO: ngram_search.c(875): bestpath 0.34 wall 0.182 xRT
l' effet lin
INFO: cmn_prior.c(131): cmn_prior_update: from < 32.09  6.52  5.60  4.29 -3.16 -8.27 -12.06 10.93  2.41 -0.40 -8.79  0.44 -0.56 >
INFO: cmn_prior.c(149): cmn_prior_update: to   < 34.37 10.57  2.93  6.10 -3.33 -8.18 -9.87  9.27  5.29 -1.59 -9.36 -0.16 -0.33 >
INFO: ngram_search_fwdtree.c(1550):     8923 words recognized (47/fr)
INFO: ngram_search_fwdtree.c(1552):   281781 senones evaluated (1483/fr)
INFO: ngram_search_fwdtree.c(1556):  2204387 channels searched (11602/fr), 105569 1st, 153123 last
INFO: ngram_search_fwdtree.c(1559):    19639 words for which last channels evaluated (103/fr)
INFO: ngram_search_fwdtree.c(1561):   259693 candidate words for entering last phone (1366/fr)
INFO: ngram_search_fwdtree.c(1564): fwdtree 12.02 CPU 6.328 xRT
INFO: ngram_search_fwdtree.c(1567): fwdtree 12.04 wall 6.335 xRT
INFO: ngram_search_fwdflat.c(302): Utterance vocabulary contains 534 words
INFO: ngram_search_fwdflat.c(948):     4119 words recognized (22/fr)
INFO: ngram_search_fwdflat.c(950):   125859 senones evaluated (662/fr)
INFO: ngram_search_fwdflat.c(952):   345149 channels searched (1816/fr)
INFO: ngram_search_fwdflat.c(954):    41363 words searched (217/fr)
INFO: ngram_search_fwdflat.c(957):    27086 word transitions (142/fr)
INFO: ngram_search_fwdflat.c(960): fwdflat 1.86 CPU 0.977 xRT
INFO: ngram_search_fwdflat.c(963): fwdflat 1.86 wall 0.978 xRT
INFO: ngram_search.c(1250): lattice start node <s>.0 end node </s>.132
INFO: ngram_search.c(1276): Eliminated 5 nodes before end node
INFO: ngram_search.c(1381): Lattice has 630 nodes, 4128 links
INFO: ps_lattice.c(1380): Bestpath score: -6334
INFO: ps_lattice.c(1384): Normalizer P(O) = alpha(</s>:132:188) = -475100
INFO: ps_lattice.c(1441): Joint P(O,S) = -491474 P(S|O) = -16374
INFO: ngram_search.c(872): bestpath 0.10 CPU 0.053 xRT
INFO: ngram_search.c(875): bestpath 0.10 wall 0.055 xRT
comme lui
INFO: ngram_search_fwdtree.c(429): TOTAL fwdtree 19.56 CPU 5.176 xRT
INFO: ngram_search_fwdtree.c(432): TOTAL fwdtree 19.58 wall 5.180 xRT
INFO: ngram_search_fwdflat.c(176): TOTAL fwdflat 4.69 CPU 1.240 xRT
INFO: ngram_search_fwdflat.c(179): TOTAL fwdflat 4.69 wall 1.242 xRT
INFO: ngram_search.c(303): TOTAL bestpath 0.44 CPU 0.117 xRT
INFO: ngram_search.c(306): TOTAL bestpath 0.45 wall 0.118 xRT

Но когда я пытаюсь использовать pocketsphinx для декодирования моих слов с микрофона, он работает на малине, но не на моем динамике.

У меня есть этот отзыв от моей малины:

pi@raspberrypi:~ $ pocketsphinx_continuous -inmic yes -adcdev plug:pcm.mic -dict sphinx-french/fr.dict -hmm sphinx-french/cmusphinx-fr-5.2 -lm sphinx-french/fr-small.lm.bin INFO: pocketsphinx.c(145): Parsed model-specific feature parameters from sphinx-french/cmusphinx-fr-5.2/feat.params Current configuration: [NAME]          [DEFLT]     [VALUE]
-agc            none        none
-agcthresh      2.0     2.000000e+00
-allphone               
-allphone_ci        no      no
-alpha          0.97        9.700000e-01
-ascale         20.0        2.000000e+01
-aw         1       1
-backtrace      no      no
-beam           1e-48       1.000000e-48
-bestpath       yes     yes
-bestpathlw     9.5     9.500000e+00
-ceplen         13      13
-cmn            current     current
-cmninit        8.0     8.0
-compallsen     no      no
-debug                  0
-dict                   sphinx-french/fr.dict
-dictcase       no      no
-dither         no      no
-doublebw       no      no
-ds         1       1
-fdict                  sphinx-french/cmusphinx-fr-5.2/noisedict
-feat           1s_c_d_dd   1s_c_d_dd
-featparams             sphinx-french/cmusphinx-fr-5.2/feat.params
-fillprob       1e-8        1.000000e-08
-frate          100     100
-fsg                    
-fsgusealtpron      yes     yes
-fsgusefiller       yes     yes
-fwdflat        yes     yes
-fwdflatbeam        1e-64       1.000000e-64
-fwdflatefwid       4       4
-fwdflatlw      8.5     8.500000e+00
-fwdflatsfwin       25      25
-fwdflatwbeam       7e-29       7.000000e-29
-fwdtree        yes     yes
-hmm                    sphinx-french/cmusphinx-fr-5.2
-input_endian       little      little
-jsgf                   
-keyphrase              
-kws                    
-kws_delay      10      10
-kws_plp        1e-1        1.000000e-01
-kws_threshold      1       1.000000e+00
-latsize        5000        5000
-lda                    sphinx-french/cmusphinx-fr-5.2/feature_transform
-ldadim         0       0
-lifter         0       22
-lm                 sphinx-french/fr-small.lm.bin
-lmctl                  
-lmname                 
-logbase        1.0001      1.000100e+00
-logfn                  
-logspec        no      no
-lowerf         133.33334   1.300000e+02
-lpbeam         1e-40       1.000000e-40
-lponlybeam     7e-29       7.000000e-29
-lw         6.5     6.500000e+00
-maxhmmpf       30000       30000
-maxwpf         -1      -1
-mdef                   sphinx-french/cmusphinx-fr-5.2/mdef
-mean                   sphinx-french/cmusphinx-fr-5.2/means
-mfclogdir              
-min_endfr      0       0
-mixw                   sphinx-french/cmusphinx-fr-5.2/mixture_weights
-mixwfloor      0.0000001   1.000000e-07
-mllr                   
-mmap           yes     yes
-ncep           13      13
-nfft           512     512
-nfilt          40      25
-nwpen          1.0     1.000000e+00
-pbeam          1e-48       1.000000e-48
-pip            1.0     1.000000e+00
-pl_beam        1e-10       1.000000e-10
-pl_pbeam       1e-10       1.000000e-10
-pl_pip         1.0     1.000000e+00
-pl_weight      3.0     3.000000e+00
-pl_window      5       5
-rawlogdir              
-remove_dc      no      no
-remove_noise       yes     yes
-remove_silence     yes     yes
-round_filters      yes     yes
-samprate       16000       1.600000e+04
-seed           -1      -1
-sendump                
-senlogdir              
-senmgau                
-silprob        0.005       5.000000e-03
-smoothspec     no      no
-svspec                 
-tmat                   sphinx-french/cmusphinx-fr-5.2/transition_matrices
-tmatfloor      0.0001      1.000000e-04
-topn           4       4
-topn_beam      0       0
-toprule                
-transform      legacy      dct
-unit_area      yes     yes
-upperf         6855.4976   6.800000e+03
-uw         1.0     1.000000e+00
-vad_postspeech     50      50
-vad_prespeech      20      20
-vad_startspeech    10      10
-vad_threshold      2.0     2.000000e+00
-var                    sphinx-french/cmusphinx-fr-5.2/variances
-varfloor       0.0001      1.000000e-04
-varnorm        no      no
-verbose        no      no
-warp_params                
-warp_type      inverse_linear  inverse_linear
-wbeam          7e-29       7.000000e-29
-wip            0.65        6.500000e-01
-wlen           0.025625    2.562500e-02

INFO: feat.c(715): Initializing feature stream to type: '1s_c_d_dd', ceplen=13, CMN='current', VARNORM='no', AGC='none' INFO: cmn.c(143): mean[0]= 12.00, mean[1..12]= 0.0 INFO: acmod.c(154): Reading linear feature transformation from sphinx-french/cmusphinx-fr-5.2/feature_transform INFO: mdef.c(518): Reading model definition: sphinx-french/cmusphinx-fr-5.2/mdef INFO: bin_mdef.c(181): Allocating 101051 * 8 bytes (789 KiB) for CD tree INFO: tmat.c(206): Reading HMM transition probability matrices: sphinx-french/cmusphinx-fr-5.2/transition_matrices INFO: acmod.c(117): Attempting to use PTM computation module INFO: ms_gauden.c(198): Reading mixture gaussian parameter: sphinx-french/cmusphinx-fr-5.2/means INFO: ms_gauden.c(292): 2108 codebook, 1 feature, size:  INFO: ms_gauden.c(294):  8x32 INFO: ms_gauden.c(198): Reading mixture gaussian parameter: sphinx-french/cmusphinx-fr-5.2/variances INFO: ms_gauden.c(292): 2108 codebook, 1 feature, size:  INFO: ms_gauden.c(294):  8x32 INFO: ms_gauden.c(354): 0 variance values floored INFO: ptm_mgau.c(801): Number of codebooks exceeds 256: 2108 INFO: acmod.c(119): Attempting to use semi-continuous computation module INFO: ms_gauden.c(198): Reading mixture gaussian parameter: sphinx-french/cmusphinx-fr-5.2/means INFO: ms_gauden.c(292): 2108 codebook, 1 feature, size:  INFO: ms_gauden.c(294):  8x32 INFO: ms_gauden.c(198): Reading mixture gaussian parameter: sphinx-french/cmusphinx-fr-5.2/variances INFO: ms_gauden.c(292): 2108 codebook, 1 feature, size:  INFO: ms_gauden.c(294):  8x32 INFO: ms_gauden.c(354): 0 variance values floored INFO: acmod.c(121): Falling back to general multi-stream GMM computation INFO: ms_gauden.c(198): Reading mixture gaussian parameter: sphinx-french/cmusphinx-fr-5.2/means INFO: ms_gauden.c(292): 2108 codebook, 1 feature, size:  INFO: ms_gauden.c(294):  8x32 INFO: ms_gauden.c(198): Reading mixture gaussian parameter: sphinx-french/cmusphinx-fr-5.2/variances INFO: ms_gauden.c(292): 2108 codebook, 1 feature, size:  INFO: ms_gauden.c(294):  8x32 INFO: ms_gauden.c(354): 0 variance values floored INFO: ms_senone.c(149): Reading senone mixture weights: sphinx-french/cmusphinx-fr-5.2/mixture_weights INFO: ms_senone.c(200): Truncating senone logs3(pdf) values by 10 bits INFO: ms_senone.c(207): Not transposing mixture weights in memory INFO: ms_senone.c(268): Read mixture weights for 2108 senones: 1 features x 8 codewords INFO: ms_senone.c(320): Mapping senones to individual codebooks INFO: ms_mgau.c(141): The value of topn: 4 INFO: phone_loop_search.c(114): State beam -225 Phone exit beam -225 Insertion penalty 0 INFO: dict.c(320): Allocating 109102 * 20 bytes (2130 KiB) for word entries INFO: dict.c(333): Reading main dictionary: sphinx-french/fr.dict INFO: dict.c(213): Allocated 1018 KiB for strings, 1375 KiB for phones INFO: dict.c(336): 105003 words read INFO: dict.c(358): Reading filler dictionary: sphinx-french/cmusphinx-fr-5.2/noisedict INFO: dict.c(213): Allocated 0 KiB for strings, 0 KiB for phones INFO: dict.c(361): 3 words read INFO: dict2pid.c(396): Building PID tables for dictionary INFO: dict2pid.c(406): Allocating 36^3 * 2 bytes (91 KiB) for word-initial triphones INFO: dict2pid.c(132): Allocated 15696 bytes (15 KiB) for word-final triphones INFO: dict2pid.c(196): Allocated 15696 bytes (15 KiB) for single-phone word triphones INFO: ngram_model_trie.c(456): Trying to read LM in trie binary format INFO: ngram_search_fwdtree.c(99): 742 unique initial diphones INFO: ngram_search_fwdtree.c(148): 0 root, 0 non-root channels, 132 single-phone words INFO: ngram_search_fwdtree.c(186): Creating search tree INFO: ngram_search_fwdtree.c(192): before: 0 root, 0 non-root channels, 132 single-phone words INFO: ngram_search_fwdtree.c(326): after: max nonroot chan increased to 122777 INFO: ngram_search_fwdtree.c(339): after: 733 root, 122649 non-root channels, 124 single-phone words INFO: ngram_search_fwdflat.c(157): fwdflat: min_ef_width = 4, max_sf_win = 25 INFO: continuous.c(305): pocketsphinx_continuous COMPILED ON: May 22 2016, AT: 22:01:16

READY.... Listening... INFO: ngram_search.c(459): Resized backpointer table to 10000 entries INFO: ngram_search.c(467): Resized score stack to 200000 entries INFO: ngram_search.c(459): Resized backpointer table to 20000 entries INFO: ngram_search.c(467): Resized score stack to 400000 entries INFO: ngram_search.c(459): Resized backpointer table to 40000 entries INFO: ngram_search_fwdtree.c(952): cand_sf[] increased to 64 entries Input overrun, read calls are too rare (non-fatal) INFO: cmn_prior.c(131): cmn_prior_update: from <  8.00  0.00  0.00  0.00 
0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00 > INFO: cmn_prior.c(149): cmn_prior_update: to   < 20.14  5.01  9.56  2.81
-0.91  5.44  1.24 -1.23 -0.46 -1.34 -0.56  1.64  3.27 > INFO: ngram_search_fwdtree.c(1553):    31150 words recognized (98/fr) INFO: ngram_search_fwdtree.c(1555):   504243 senones evaluated (1591/fr) INFO: ngram_search_fwdtree.c(1559):  6306963 channels searched (19895/fr), 208020 1st, 615812 last INFO: ngram_search_fwdtree.c(1562):    61696 words for which last channels evaluated (194/fr) INFO: ngram_search_fwdtree.c(1564):   787418 candidate words for entering last phone (2483/fr) INFO: ngram_search_fwdtree.c(1567): fwdtree 21.23 CPU 6.698 xRT INFO: ngram_search_fwdtree.c(1570): fwdtree 29.45 wall 9.290 xRT INFO: ngram_search_fwdflat.c(302): Utterance vocabulary contains 1051 words INFO: ngram_search_fwdflat.c(948):    16647 words recognized (53/fr) INFO: ngram_search_fwdflat.c(950):   355200 senones evaluated (1121/fr) INFO: ngram_search_fwdflat.c(952):  1666450 channels searched (5256/fr) INFO: ngram_search_fwdflat.c(954):   142520 words searched (449/fr) INFO: ngram_search_fwdflat.c(957):    71616 word transitions (225/fr) INFO: ngram_search_fwdflat.c(960): fwdflat 6.84 CPU 2.158 xRT INFO: ngram_search_fwdflat.c(963): fwdflat 6.84 wall
2.159 xRT INFO: ngram_search.c(1200): </s> not found in last frame, using membres.315 instead INFO: ngram_search.c(1253): lattice start node <s>.0 end node membres.233 INFO: ngram_search.c(1279): Eliminated 201 nodes before end node INFO: ngram_search.c(1384): Lattice has 1802 nodes, 40836 links INFO: ps_lattice.c(1380): Bestpath score: -10518 INFO: ps_lattice.c(1384): Normalizer P(O) = alpha(membres:233:315) =
-772446 INFO: ps_lattice.c(1441): Joint P(O,S) = -795829 P(S|O) = -23383 INFO: ngram_search.c(875): bestpath 3.10 CPU 0.982 xRT INFO: ngram_search.c(878): bestpath 3.10 wall 0.982 xRT on vous vous vous membres

Но все заканчивается этим откликом на моем динамике (я не выкладываю весь журнал из-за ограничения символов):

INFO: continuous.c(252): Ready....                                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)                              
Input overrun, read calls are too rare (non-fatal)

(Для заметки я используюфранцузская модель карманного сфинкса)

Мой вопрос: есть ли у вас идеи почему?Вероятно, это не проблема обработки энергии, поскольку она работает на малине.

1 Ответ

0 голосов
/ 31 мая 2019

В сообщении говорится, что обработка выполняется слишком медленно.

Вам нужно запустить декодирование из файла на вашем устройстве, а не из микрофона, и сначала настроить скорость. Скорость должна быть быстрее, чем в реальном времени в журналах.

Вам необходимо использовать меньшие модели или оптимизировать скорость с помощью лучей / других параметров, как описано в вики

...