CMU sphinx训练命令控制声学模型问题
2012-07-31 11:48
393 查看
(此片文章适用于训练命令控制的声学模型过程中遇到问题的朋友)
开始学习CMU的sphinx，感觉做语音的好苦逼啊，至少对于我来说。
从网上找到一个不错的教程：Sphinx武林秘籍，我已经转过来了，这篇文章里就不在赘述了。
按照“武林秘籍”方法，我尝试构建自己的建议的语音命令控制系统，利用这个过程了解一下SPHINX。然而，在训练自己的声学模型时候，完全无法训练成功。或者说，按照那种方式，训练模型的操作根本没有开始：其数据显示如下：
MODULE: 00 verify training files
O.S. is case sensitive ("A" != "a").
Phones will be treated as case sensitive.
Phase 1: DICT - Checking to see if the dict and filler dict agrees with the phonelist file.
Found 8 words using 16 phones
Phase 2: DICT - Checking to make sure there are not duplicate entries in the dictionary
Phase 3: CTL - Check general format; utterance length (must be positive); files exist
Phase 4: CTL - Checking number of lines in the transcript should match lines in control file
Phase 5: CTL - Determine amount of training data, see if n_tied_states seems reasonable.
Estimated Total Hours Training: 0.00230277777777778
This is a small amount of data, no comment at this time
Phase 6: TRANSCRIPT - Checking that all the words in the transcript are in the dictionary
Words in dictionary: 5
Words in filler dictionary: 3
Phase 7: TRANSCRIPT - Checking that all the phones in the transcript are in the phonelist, and all phones in the phonelist appear at least once
MODULE: 01 Train LDA transformation
Skipped (set $CFG_LDA_MLLT = 'yes' to enable)
MODULE: 02 Train MLLT transformation
Skipped (set $CFG_LDA_MLLT = 'yes' to enable)
MODULE: 05 Vector Quantization
Skipped for continuous models
MODULE: 10 Training Context Independent models for forced alignment and VTLN
Skipped: $ST::CFG_FORCEDALIGN set to 'no' in sphinx_train.cfg
Skipped: $ST::CFG_VTLN set to 'no' in sphinx_train.cfg
MODULE: 11 Force-aligning transcripts
Skipped: $ST::CFG_FORCEDALIGN set to 'no' in sphinx_train.cfg
MODULE: 12 Force-aligning data for VTLN
Skipped: $ST::CFG_VTLN set to 'no' in sphinx_train.cfg
MODULE: 20 Training Context Independent models
Phase 1: Cleaning up directories:
accumulator...logs...qmanager...models...
Phase 2: Flat initialize
Phase 3: Forward-Backward
Baum welch starting for 1 Gaussian(s), iteration: 1 (1 of 1)
0%
Normalization for iteration: 1
Current Overall Likelihood Per Frame = -2.70601326899879
Baum welch starting for 1 Gaussian(s), iteration: 2 (1 of 1)
0%
Normalization for iteration: 2
Current Overall Likelihood Per Frame = 5.81870808202654
Convergence Ratio = 8.52472135102533
Baum welch starting for 1 Gaussian(s), iteration: 3 (1 of 1)
0%
Normalization for iteration: 3
Current Overall Likelihood Per Frame = 12.7354402895054
Convergence Ratio = 6.91673220747889
Baum welch starting for 1 Gaussian(s), iteration: 4 (1 of 1)
0%
Normalization for iteration: 4
Current Overall Likelihood Per Frame = 16.0007961399276
Convergence Ratio = 3.26535585042222
Baum welch starting for 1 Gaussian(s), iteration: 5 (1 of 1)
0%
Normalization for iteration: 5
Current Overall Likelihood Per Frame = 16.3007720144753
Convergence Ratio = 0.29997587454767
Baum welch starting for 1 Gaussian(s), iteration: 6 (1 of 1)
0%
Normalization for iteration: 6
Current Overall Likelihood Per Frame = 16.3483956574186
Training completed after 6 iterations
MODULE: 30 Training Context Dependent models
Phase 1: Cleaning up directories:
accumulator...logs...qmanager...
Phase 2: Initialization
Phase 3: Forward-Backward
Baum welch starting for iteration: 1 (1 of 1)
0%
Normalization for iteration: 1
Current Overall Likelihood Per Frame = 16.3717370325694
Baum welch starting for iteration: 2 (1 of 1)
0%
Normalization for iteration: 2
Current Overall Likelihood Per Frame = 17.507056694813
Training completed after 2 iterations
MODULE: 40 Build Trees
Phase 1: Cleaning up old log files...
Phase 2: Make Questions
Phase 3: Tree building
Processing each phone with each state
H 0
H 1
H 2
IAN 0
IAN 1
IAN 2
IB 0
IB 1
IB 2
IN 0
IN 1
IN 2
ING 0
ING 1
ING 2
J 0
J 1
J 2
OU 0
OU 1
OU 2
Q 0
Q 1
Q 2
Skipping SIL
T 0
T 1
T 2
UAN 0
UAN 1
UAN 2
UI 0
UI 1
UI 2
UO 0
UO 1
UO 2
Y 0
Y 1
Y 2
Z 0
Z 1
Z 2
ZH 0
ZH 1
ZH 2
MODULE: 45 Prune Trees
Phase 1: Tree Pruning
WARNING: This step had 0 ERROR messages and 1 WARNING messages. Please check the log file for details.
Phase 2: State Tying
MODULE: 50 Training Context dependent models
Phase 1: Cleaning up directories:
accumulator...logs...qmanager...
Phase 2: Copy CI to CD initialize
Phase 3: Forward-Backward
Baum welch starting for 1 Gaussian(s), iteration: 1 (1 of 1)
0%
Normalization for iteration: 1
This step had 132 ERROR messages and 0 WARNING messages. Please check the log file for details.
Current Overall Likelihood Per Frame = 16.3717370325694
Baum welch starting for 1 Gaussian(s), iteration: 2 (1 of 1)
0%
Normalization for iteration: 2
Current Overall Likelihood Per Frame = 18.1674185765983
Convergence Ratio = 1.79568154402891
Baum welch starting for 1 Gaussian(s), iteration: 3 (1 of 1)
0%
Normalization for iteration: 3
Current Overall Likelihood Per Frame = 18.1774788902292
Split Gaussians, increase by 1
Current Overall Likelihood Per Frame = 18.1774788902292
Convergence Ratio = 0.0100603136308912
Baum welch starting for 2 Gaussian(s), iteration: 1 (1 of 1)
0%
Normalization for iteration: 1
This step had 264 ERROR messages and 0 WARNING messages. Please check the log file for details.
Current Overall Likelihood Per Frame = 17.7069481302774
Baum welch starting for 2 Gaussian(s), iteration: 2 (1 of 1)
0%
Normalization for iteration: 2
Current Overall Likelihood Per Frame = 20.5196019300362
Convergence Ratio = 2.81265379975879
Baum welch starting for 2 Gaussian(s), iteration: 3 (1 of 1)
0%
Normalization for iteration: 3
Current Overall Likelihood Per Frame = 23.2946200241255
Convergence Ratio = 2.77501809408925
Baum welch starting for 2 Gaussian(s), iteration: 4 (1 of 1)
0%
Normalization for iteration: 4
Current Overall Likelihood Per Frame = 25.648902291918
Convergence Ratio = 2.35428226779247
Baum welch starting for 2 Gaussian(s), iteration: 5 (1 of 1)
0%
Normalization for iteration: 5
Current Overall Likelihood Per Frame = 26.339312424608
Convergence Ratio = 0.690410132689962
Baum welch starting for 2 Gaussian(s), iteration: 6 (1 of 1)
0%
Normalization for iteration: 6
Current Overall Likelihood Per Frame = 26.4127744270205
Split Gaussians, increase by 2
Current Overall Likelihood Per Frame = 26.4127744270205
Convergence Ratio = 0.073462002412505
Baum welch starting for 4 Gaussian(s), iteration: 1 (1 of 1)
0%
Normalization for iteration: 1
This step had 1635 ERROR messages and 0 WARNING messages. Please check the log file for details.
Current Overall Likelihood Per Frame = 25.9812545235223
Baum welch starting for 4 Gaussian(s), iteration: 2 (1 of 1)
0%
Normalization for iteration: 2
Current Overall Likelihood Per Frame = 29.3810373944511
Convergence Ratio = 3.39978287092885
Baum welch starting for 4 Gaussian(s), iteration: 3 (1 of 1)
0%
Normalization for iteration: 3
Current Overall Likelihood Per Frame = 33.7967671893848
Convergence Ratio = 4.4157297949337
Baum welch starting for 4 Gaussian(s), iteration: 4 (1 of 1)
0%
Normalization for iteration: 4
Current Overall Likelihood Per Frame = 39.2286248492159
Convergence Ratio = 5.43185765983112
Baum welch starting for 4 Gaussian(s), iteration: 5 (1 of 1)
0%
Normalization for iteration: 5
Current Overall Likelihood Per Frame = 39.9667068757539
Convergence Ratio = 0.738082026538024
Baum welch starting for 4 Gaussian(s), iteration: 6 (1 of 1)
0%
Normalization for iteration: 6
Current Overall Likelihood Per Frame = 40.0017852834741
Split Gaussians, increase by 4
Current Overall Likelihood Per Frame = 40.0017852834741
Convergence Ratio = 0.0350784077201709
Baum welch starting for 8 Gaussian(s), iteration: 1 (1 of 1)
0%
Normalization for iteration: 1
This step had 6941 ERROR messages and 0 WARNING messages. Please check the log file for details.
Current Overall Likelihood Per Frame = 39.5515681544029
Baum welch starting for 8 Gaussian(s), iteration: 2 (1 of 1)
0%
Normalization for iteration: 2
Current Overall Likelihood Per Frame = 43.9104342581423
Convergence Ratio = 4.35886610373944
Baum welch starting for 8 Gaussian(s), iteration: 3 (1 of 1)
0%
Normalization for iteration: 3
Current Overall Likelihood Per Frame = 50.497189384801
Convergence Ratio = 6.58675512665866
Baum welch starting for 8 Gaussian(s), iteration: 4 (1 of 1)
0%
Normalization for iteration: 4
Current Overall Likelihood Per Frame = 62.2500844390832
Convergence Ratio = 11.7528950542822
Baum welch starting for 8 Gaussian(s), iteration: 5 (1 of 1)
0%
Normalization for iteration: 5
Current Overall Likelihood Per Frame = 63.5334861278649
Convergence Ratio = 1.28340168878169
Baum welch starting for 8 Gaussian(s), iteration: 6 (1 of 1)
0%
Normalization for iteration: 6
Current Overall Likelihood Per Frame = 63.6126658624849
Split Gaussians, increase by 0
Training for 8 Gaussian(s) completed after 6 iterations
MODULE: 60 Lattice Generation
Skipped: $ST::CFG_MMIE set to 'no' in sphinx_train.cfg
MODULE: 61 Lattice Pruning
Skipped: $ST::CFG_MMIE set to 'no' in sphinx_train.cfg
MODULE: 62 Lattice Format Conversion
Skipped: $ST::CFG_MMIE set to 'no' in sphinx_train.cfg
MODULE: 65 MMIE Training
Skipped: $ST::CFG_MMIE set to 'no' in sphinx_train.cfg
MODULE: 90 deleted interpolation
Skipped for continuous models
这个问题纠结了很久，从0%可以看出，训练根本就没有启动，也查看了SPHINX的官方指导：http://cmusphinx.sourceforge.net/wiki/tutorialam。仍然不能的到解决。不过最终发现了问题所在：训练数据不足。官方虽然明确“1 hour of recording for command and control for single speaker”，但是我看“武林秘籍”作者既然可以训练出来，那么这里数据太少应该不适问题，故没有重视。但是经过N天的折腾之后，依然毫无改善，就去尝试增加训练数据，看看结果，没想到竟然真的是这里的问题。
我的训练数据离1hour还很远，只有“Estimated Total Hours Training: 0.0633277777777778”，但是相对于每个命令只录制一遍，有了相当大的改善（至少声学模型训练可以成功，而不是满天的ERROR）。之后将尝试继续添加训练数据，看看效果能否继续改善。
对于：Warning: Could not find Mic element这个问题，由于我是使用的Ubuntu的Wubi版，不知道和这个有没有关系，直接采取忽略的做法，原因当然是我说话，它能：
READY....
Listening...
Recording is stopped, start recording with ad_start_rec。
最终结果还不是很理想，“前进”偶尔会识别为“停止”，“右转”基本都识别为“停止”，估计原因是录制的音频不是很自然。之后将尝试添加更多训练数据以期改善。不过毕竟有了结果，这还是令人欣慰的。
运行结果如下：
$:pocketsphinx_continuous -hmm my_db.cd_cont_100 -lm my_db.lm.DMP -dict my_db.dic
INFO: cmd_ln.c(691): Parsing command line:
pocketsphinx_continuous \
-hmm my_db.cd_cont_100 \
-lm my_db.lm.DMP \
-dict my_db.dic
Current configuration:
[NAME] [DEFLT] [VALUE]
-adcdev
-agc none none
-agcthresh 2.0 2.000000e+00
-alpha 0.97 9.700000e-01
-argfile
-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
-bghist no no
-ceplen 13 13
-cmn current current
-cmninit 8.0 8.0
-compallsen no no
-debug 0
-dict my_db.dic
-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 my_db.cd_cont_100
-infile
-input_endian little little
-jsgf
-kdmaxbbi -1 -1
-kdmaxdepth 0 0
-kdtree
-latsize 5000 5000
-lda
-ldadim 0 0
-lextreedump 0 0
-lifter 0 0
-lm my_db.lm.DMP
-lmctl
-lmname default default
-logbase 1.0001 1.000100e+00
-logfn
-logspec no no
-lowerf 133.33334 1.333333e+02
-lpbeam 1e-40 1.000000e-40
-lponlybeam 7e-29 7.000000e-29
-lw 6.5 6.500000e+00
-maxhmmpf -1 -1
-maxnewoov 20 20
-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 40
-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-5 1.000000e-05
-pl_window 0 0
-rawlogdir
-remove_dc no no
-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
-time no no
-tmat
-tmatfloor 0.0001 1.000000e-04
-topn 4 4
-topn_beam 0 0
-toprule
-transform legacy legacy
-unit_area yes yes
-upperf 6855.4976 6.855498e+03
-usewdphones no no
-uw 1.0 1.000000e+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: cmd_ln.c(691): Parsing command line:
\
-alpha 0.97 \
-doublebw no \
-nfilt 40 \
-ncep 13 \
-lowerf 133.33334 \
-upperf 6855.4976 \
-nfft 512 \
-wlen 0.0256 \
-transform legacy \
-feat 1s_c_d_dd \
-agc none \
-cmn current \
-varnorm no
Current configuration:
[NAME] [DEFLT] [VALUE]
-agc none none
-agcthresh 2.0 2.000000e+00
-alpha 0.97 9.700000e-01
-ceplen 13 13
-cmn current current
-cmninit 8.0 8.0
-dither no no
-doublebw no no
-feat 1s_c_d_dd 1s_c_d_dd
-frate 100 100
-input_endian little little
-lda
-ldadim 0 0
-lifter 0 0
-logspec no no
-lowerf 133.33334 1.333333e+02
-ncep 13 13
-nfft 512 512
-nfilt 40 40
-remove_dc no no
-round_filters yes yes
-samprate 16000 1.600000e+04
-seed -1 -1
-smoothspec no no
-svspec
-transform legacy legacy
-unit_area yes yes
-upperf 6855.4976 6.855498e+03
-varnorm no no
-verbose no no
-warp_params
-warp_type inverse_linear inverse_linear
-wlen 0.025625 2.560000e-02
INFO: acmod.c(242): Parsed model-specific feature parameters from my_db.cd_cont_100/feat.params
INFO: feat.c(684): Initializing feature stream to type: '1s_c_d_dd', ceplen=13, CMN='current', VARNORM='no', AGC='none'
INFO: cmn.c(142): mean[0]= 12.00, mean[1..12]= 0.0
INFO: mdef.c(520): Reading model definition: my_db.cd_cont_100/mdef
INFO: bin_mdef.c(173): Allocating 166 * 8 bytes (1 KiB) for CD tree
INFO: tmat.c(205): Reading HMM transition probability matrices: my_db.cd_cont_100/transition_matrices
INFO: acmod.c(117): Attempting to use SCHMM computation module
INFO: ms_gauden.c(198): Reading mixture gaussian parameter: my_db.cd_cont_100/means
INFO: ms_gauden.c(292): 105 codebook, 1 feature, size:
INFO: ms_gauden.c(294): 8x39
INFO: ms_gauden.c(198): Reading mixture gaussian parameter: my_db.cd_cont_100/variances
INFO: ms_gauden.c(292): 105 codebook, 1 feature, size:
INFO: ms_gauden.c(294): 8x39
INFO: ms_gauden.c(354): 412 variance values floored
INFO: acmod.c(119): Attempting to use PTHMM computation module
INFO: ms_gauden.c(198): Reading mixture gaussian parameter: my_db.cd_cont_100/means
INFO: ms_gauden.c(292): 105 codebook, 1 feature, size:
INFO: ms_gauden.c(294): 8x39
INFO: ms_gauden.c(198): Reading mixture gaussian parameter: my_db.cd_cont_100/variances
INFO: ms_gauden.c(292): 105 codebook, 1 feature, size:
INFO: ms_gauden.c(294): 8x39
INFO: ms_gauden.c(354): 412 variance values floored
INFO: ptm_mgau.c(804): Number of codebooks doesn't match number of ciphones, doesn't look like PTM: 105 16
INFO: acmod.c(121): Falling back to general multi-stream GMM computation
INFO: ms_gauden.c(198): Reading mixture gaussian parameter: my_db.cd_cont_100/means
INFO: ms_gauden.c(292): 105 codebook, 1 feature, size:
INFO: ms_gauden.c(294): 8x39
INFO: ms_gauden.c(198): Reading mixture gaussian parameter: my_db.cd_cont_100/variances
INFO: ms_gauden.c(292): 105 codebook, 1 feature, size:
INFO: ms_gauden.c(294): 8x39
INFO: ms_gauden.c(354): 412 variance values floored
INFO: ms_senone.c(160): Reading senone mixture weights: my_db.cd_cont_100/mixture_weights
INFO: ms_senone.c(211): Truncating senone logs3(pdf) values by 10 bits
INFO: ms_senone.c(218): Not transposing mixture weights in memory
INFO: ms_senone.c(277): Read mixture weights for 105 senones: 1 features x 8 codewords
INFO: ms_senone.c(331): Mapping senones to individual codebooks
INFO: ms_mgau.c(122): The value of topn: 4
INFO: dict.c(306): Allocating 4104 * 20 bytes (80 KiB) for word entries
INFO: dict.c(321): Reading main dictionary: my_db.dic
INFO: dict.c(212): Allocated 0 KiB for strings, 0 KiB for phones
INFO: dict.c(324): 5 words read
INFO: dict.c(330): Reading filler dictionary: my_db.cd_cont_100/noisedict
INFO: dict.c(212): Allocated 0 KiB for strings, 0 KiB for phones
INFO: dict.c(333): 3 words read
INFO: dict2pid.c(396): Building PID tables for dictionary
INFO: dict2pid.c(404): Allocating 16^3 * 2 bytes (8 KiB) for word-initial triphones
INFO: dict2pid.c(131): Allocated 3136 bytes (3 KiB) for word-final triphones
INFO: dict2pid.c(195): Allocated 3136 bytes (3 KiB) for single-phone word triphones
INFO: ngram_model_arpa.c(77): No \data\ mark in LM file
INFO: ngram_model_dmp.c(142): Will use memory-mapped I/O for LM file
INFO: ngram_model_dmp.c(196): ngrams 1=7, 2=10, 3=13
INFO: ngram_model_dmp.c(242): 7 = LM.unigrams(+trailer) read
INFO: ngram_model_dmp.c(291): 10 = LM.bigrams(+trailer) read
INFO: ngram_model_dmp.c(317): 13 = LM.trigrams read
INFO: ngram_model_dmp.c(342): 4 = LM.prob2 entries read
INFO: ngram_model_dmp.c(362): 5 = LM.bo_wt2 entries read
INFO: ngram_model_dmp.c(382): 3 = LM.prob3 entries read
INFO: ngram_model_dmp.c(410): 1 = LM.tseg_base entries read
INFO: ngram_model_dmp.c(466): 7 = ascii word strings read
INFO: ngram_search_fwdtree.c(99): 5 unique initial diphones
INFO: ngram_search_fwdtree.c(147): 0 root, 0 non-root channels, 4 single-phone words
INFO: ngram_search_fwdtree.c(186): Creating search tree
INFO: ngram_search_fwdtree.c(191): before: 0 root, 0 non-root channels, 4 single-phone words
INFO: ngram_search_fwdtree.c(326): after: max nonroot chan increased to 138
INFO: ngram_search_fwdtree.c(338): after: 5 root, 10 non-root channels, 3 single-phone words
INFO: ngram_search_fwdflat.c(156): fwdflat: min_ef_width = 4, max_sf_win = 25
INFO: continuous.c(367): pocketsphinx_continuous COMPILED ON: Jul 29 2012, AT: 18:16:08
Warning: Could not find Mic element
READY....
Listening...
Recording is stopped, start recording with ad_start_rec
Stopped listening, please wait...
INFO: cmn_prior.c(121): 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(139): cmn_prior_update: to < 9.71 1.10 -0.35 -0.07 -0.45 -0.21 -0.27 0.05 -0.21 -0.26 -0.14 -0.11 -0.06 >
INFO: ngram_search_fwdtree.c(1549): 88 words recognized (1/fr)
INFO: ngram_search_fwdtree.c(1551): 1416 senones evaluated (16/fr)
INFO: ngram_search_fwdtree.c(1553): 548 channels searched (6/fr), 203 1st, 171 last
INFO: ngram_search_fwdtree.c(1557): 171 words for which last channels evaluated (1/fr)
INFO: ngram_search_fwdtree.c(1560): 38 candidate words for entering last phone (0/fr)
INFO: ngram_search_fwdtree.c(1562): fwdtree 0.01 CPU 0.014 xRT
INFO: ngram_search_fwdtree.c(1565): fwdtree 1.40 wall 1.632 xRT
INFO: ngram_search_fwdflat.c(305): Utterance vocabulary contains 3 words
INFO: ngram_search_fwdflat.c(940): 68 words recognized (1/fr)
INFO: ngram_search_fwdflat.c(942): 912 senones evaluated (11/fr)
INFO: ngram_search_fwdflat.c(944): 384 channels searched (4/fr)
INFO: ngram_search_fwdflat.c(946): 209 words searched (2/fr)
INFO: ngram_search_fwdflat.c(948): 101 word transitions (1/fr)
INFO: ngram_search_fwdflat.c(951): fwdflat -0.00 CPU -0.000 xRT
INFO: ngram_search_fwdflat.c(954): fwdflat 0.00 wall 0.003 xRT
INFO: ngram_search.c(1253): lattice start node <s>.0 end node </s>.83
INFO: ngram_search.c(1281): Eliminated 0 nodes before end node
INFO: ngram_search.c(1386): Lattice has 10 nodes, 8 links
INFO: ps_lattice.c(1352): Normalizer P(O) = alpha(</s>:83:84) = -258409
INFO: ps_lattice.c(1390): Joint P(O,S) = -258409 P(S|O) = 0
INFO: ngram_search.c(875): bestpath 0.00 CPU 0.000 xRT
INFO: ngram_search.c(878): bestpath 0.00 wall 0.000 xRT
000000000: 停止
READY....
Listening...
Recording is stopped, start recording with ad_start_rec
Stopped listening, please wait...
INFO: cmn_prior.c(121): cmn_prior_update: from < 9.71 1.10 -0.35 -0.07 -0.45 -0.21 -0.27 0.05 -0.21 -0.26 -0.14 -0.11 -0.06 >
INFO: cmn_prior.c(139): cmn_prior_update: to < 9.75 1.06 -0.31 -0.03 -0.41 -0.20 -0.24 0.06 -0.20 -0.28 -0.13 -0.14 -0.08 >
INFO: ngram_search_fwdtree.c(1549): 154 words recognized (2/fr)
INFO: ngram_search_fwdtree.c(1551): 1848 senones evaluated (23/fr)
INFO: ngram_search_fwdtree.c(1553): 718 channels searched (8/fr), 309 1st, 193 last
INFO: ngram_search_fwdtree.c(1557): 193 words for which last channels evaluated (2/fr)
INFO: ngram_search_fwdtree.c(1560): 8 candidate words for entering last phone (0/fr)
INFO: ngram_search_fwdtree.c(1562): fwdtree 0.02 CPU 0.020 xRT
INFO: ngram_search_fwdtree.c(1565): fwdtree 2.88 wall 3.601 xRT
INFO: ngram_search_fwdflat.c(305): Utterance vocabulary contains 3 words
INFO: ngram_search_fwdflat.c(940): 87 words recognized (1/fr)
INFO: ngram_search_fwdflat.c(942): 933 senones evaluated (12/fr)
INFO: ngram_search_fwdflat.c(944): 427 channels searched (5/fr)
INFO: ngram_search_fwdflat.c(946): 256 words searched (3/fr)
INFO: ngram_search_fwdflat.c(948): 120 word transitions (1/fr)
INFO: ngram_search_fwdflat.c(951): fwdflat 0.00 CPU 0.000 xRT
INFO: ngram_search_fwdflat.c(954): fwdflat 0.00 wall 0.003 xRT
INFO: ngram_search.c(1253): lattice start node <s>.0 end node </s>.71
INFO: ngram_search.c(1281): Eliminated 0 nodes before end node
INFO: ngram_search.c(1386): Lattice has 12 nodes, 9 links
INFO: ps_lattice.c(1352): Normalizer P(O) = alpha(</s>:71:78) = -227997
INFO: ps_lattice.c(1390): Joint P(O,S) = -227997 P(S|O) = 0
INFO: ngram_search.c(875): bestpath -0.00 CPU -0.000 xRT
INFO: ngram_search.c(878): bestpath 0.00 wall 0.000 xRT
000000001: 停止
READY....
Listening...
Recording is stopped, start recording with ad_start_rec
Stopped listening, please wait...
INFO: cmn_prior.c(121): cmn_prior_update: from < 9.75 1.06 -0.31 -0.03 -0.41 -0.20 -0.24 0.06 -0.20 -0.28 -0.13 -0.14 -0.08 >
INFO: cmn_prior.c(139): cmn_prior_update: to < 9.84 1.08 -0.32 -0.03 -0.43 -0.22 -0.24 0.04 -0.20 -0.27 -0.15 -0.13 -0.09 >
INFO: ngram_search_fwdtree.c(1549): 137 words recognized (1/fr)
INFO: ngram_search_fwdtree.c(1551): 2142 senones evaluated (23/fr)
INFO: ngram_search_fwdtree.c(1553): 801 channels searched (8/fr), 313 1st, 228 last
INFO: ngram_search_fwdtree.c(1557): 228 words for which last channels evaluated (2/fr)
INFO: ngram_search_fwdtree.c(1560): 56 candidate words for entering last phone (0/fr)
INFO: ngram_search_fwdtree.c(1562): fwdtree 0.02 CPU 0.017 xRT
INFO: ngram_search_fwdtree.c(1565): fwdtree 1.38 wall 1.487 xRT
INFO: ngram_search_fwdflat.c(305): Utterance vocabulary contains 4 words
INFO: ngram_search_fwdflat.c(940): 108 words recognized (1/fr)
INFO: ngram_search_fwdflat.c(942): 1455 senones evaluated (16/fr)
INFO: ngram_search_fwdflat.c(944): 590 channels searched (6/fr)
INFO: ngram_search_fwdflat.c(946): 351 words searched (3/fr)
INFO: ngram_search_fwdflat.c(948): 154 word transitions (1/fr)
INFO: ngram_search_fwdflat.c(951): fwdflat 0.00 CPU 0.000 xRT
INFO: ngram_search_fwdflat.c(954): fwdflat 0.00 wall 0.004 xRT
INFO: ngram_search.c(1253): lattice start node <s>.0 end node </s>.90
INFO: ngram_search.c(1281): Eliminated 0 nodes before end node
INFO: ngram_search.c(1386): Lattice has 13 nodes, 12 links
INFO: ps_lattice.c(1352): Normalizer P(O) = alpha(</s>:90:91) = -208439
INFO: ps_lattice.c(1390): Joint P(O,S) = -208439 P(S|O) = 0
INFO: ngram_search.c(875): bestpath -0.00 CPU -0.000 xRT
INFO: ngram_search.c(878): bestpath 0.00 wall 0.000 xRT
000000002: 左转
READY....
开始学习CMU的sphinx，感觉做语音的好苦逼啊，至少对于我来说。
从网上找到一个不错的教程：Sphinx武林秘籍，我已经转过来了，这篇文章里就不在赘述了。
按照“武林秘籍”方法，我尝试构建自己的建议的语音命令控制系统，利用这个过程了解一下SPHINX。然而，在训练自己的声学模型时候，完全无法训练成功。或者说，按照那种方式，训练模型的操作根本没有开始：其数据显示如下：
MODULE: 00 verify training files
O.S. is case sensitive ("A" != "a").
Phones will be treated as case sensitive.
Phase 1: DICT - Checking to see if the dict and filler dict agrees with the phonelist file.
Found 8 words using 16 phones
Phase 2: DICT - Checking to make sure there are not duplicate entries in the dictionary
Phase 3: CTL - Check general format; utterance length (must be positive); files exist
Phase 4: CTL - Checking number of lines in the transcript should match lines in control file
Phase 5: CTL - Determine amount of training data, see if n_tied_states seems reasonable.
Estimated Total Hours Training: 0.00230277777777778
This is a small amount of data, no comment at this time
Phase 6: TRANSCRIPT - Checking that all the words in the transcript are in the dictionary
Words in dictionary: 5
Words in filler dictionary: 3
Phase 7: TRANSCRIPT - Checking that all the phones in the transcript are in the phonelist, and all phones in the phonelist appear at least once
MODULE: 01 Train LDA transformation
Skipped (set $CFG_LDA_MLLT = 'yes' to enable)
MODULE: 02 Train MLLT transformation
Skipped (set $CFG_LDA_MLLT = 'yes' to enable)
MODULE: 05 Vector Quantization
Skipped for continuous models
MODULE: 10 Training Context Independent models for forced alignment and VTLN
Skipped: $ST::CFG_FORCEDALIGN set to 'no' in sphinx_train.cfg
Skipped: $ST::CFG_VTLN set to 'no' in sphinx_train.cfg
MODULE: 11 Force-aligning transcripts
Skipped: $ST::CFG_FORCEDALIGN set to 'no' in sphinx_train.cfg
MODULE: 12 Force-aligning data for VTLN
Skipped: $ST::CFG_VTLN set to 'no' in sphinx_train.cfg
MODULE: 20 Training Context Independent models
Phase 1: Cleaning up directories:
accumulator...logs...qmanager...models...
Phase 2: Flat initialize
Phase 3: Forward-Backward
Baum welch starting for 1 Gaussian(s), iteration: 1 (1 of 1)
0%
Normalization for iteration: 1
Current Overall Likelihood Per Frame = -2.70601326899879
Baum welch starting for 1 Gaussian(s), iteration: 2 (1 of 1)
0%
Normalization for iteration: 2
Current Overall Likelihood Per Frame = 5.81870808202654
Convergence Ratio = 8.52472135102533
Baum welch starting for 1 Gaussian(s), iteration: 3 (1 of 1)
0%
Normalization for iteration: 3
Current Overall Likelihood Per Frame = 12.7354402895054
Convergence Ratio = 6.91673220747889
Baum welch starting for 1 Gaussian(s), iteration: 4 (1 of 1)
0%
Normalization for iteration: 4
Current Overall Likelihood Per Frame = 16.0007961399276
Convergence Ratio = 3.26535585042222
Baum welch starting for 1 Gaussian(s), iteration: 5 (1 of 1)
0%
Normalization for iteration: 5
Current Overall Likelihood Per Frame = 16.3007720144753
Convergence Ratio = 0.29997587454767
Baum welch starting for 1 Gaussian(s), iteration: 6 (1 of 1)
0%
Normalization for iteration: 6
Current Overall Likelihood Per Frame = 16.3483956574186
Training completed after 6 iterations
MODULE: 30 Training Context Dependent models
Phase 1: Cleaning up directories:
accumulator...logs...qmanager...
Phase 2: Initialization
Phase 3: Forward-Backward
Baum welch starting for iteration: 1 (1 of 1)
0%
Normalization for iteration: 1
Current Overall Likelihood Per Frame = 16.3717370325694
Baum welch starting for iteration: 2 (1 of 1)
0%
Normalization for iteration: 2
Current Overall Likelihood Per Frame = 17.507056694813
Training completed after 2 iterations
MODULE: 40 Build Trees
Phase 1: Cleaning up old log files...
Phase 2: Make Questions
Phase 3: Tree building
Processing each phone with each state
H 0
H 1
H 2
IAN 0
IAN 1
IAN 2
IB 0
IB 1
IB 2
IN 0
IN 1
IN 2
ING 0
ING 1
ING 2
J 0
J 1
J 2
OU 0
OU 1
OU 2
Q 0
Q 1
Q 2
Skipping SIL
T 0
T 1
T 2
UAN 0
UAN 1
UAN 2
UI 0
UI 1
UI 2
UO 0
UO 1
UO 2
Y 0
Y 1
Y 2
Z 0
Z 1
Z 2
ZH 0
ZH 1
ZH 2
MODULE: 45 Prune Trees
Phase 1: Tree Pruning
WARNING: This step had 0 ERROR messages and 1 WARNING messages. Please check the log file for details.
Phase 2: State Tying
MODULE: 50 Training Context dependent models
Phase 1: Cleaning up directories:
accumulator...logs...qmanager...
Phase 2: Copy CI to CD initialize
Phase 3: Forward-Backward
Baum welch starting for 1 Gaussian(s), iteration: 1 (1 of 1)
0%
Normalization for iteration: 1
This step had 132 ERROR messages and 0 WARNING messages. Please check the log file for details.
Current Overall Likelihood Per Frame = 16.3717370325694
Baum welch starting for 1 Gaussian(s), iteration: 2 (1 of 1)
0%
Normalization for iteration: 2
Current Overall Likelihood Per Frame = 18.1674185765983
Convergence Ratio = 1.79568154402891
Baum welch starting for 1 Gaussian(s), iteration: 3 (1 of 1)
0%
Normalization for iteration: 3
Current Overall Likelihood Per Frame = 18.1774788902292
Split Gaussians, increase by 1
Current Overall Likelihood Per Frame = 18.1774788902292
Convergence Ratio = 0.0100603136308912
Baum welch starting for 2 Gaussian(s), iteration: 1 (1 of 1)
0%
Normalization for iteration: 1
This step had 264 ERROR messages and 0 WARNING messages. Please check the log file for details.
Current Overall Likelihood Per Frame = 17.7069481302774
Baum welch starting for 2 Gaussian(s), iteration: 2 (1 of 1)
0%
Normalization for iteration: 2
Current Overall Likelihood Per Frame = 20.5196019300362
Convergence Ratio = 2.81265379975879
Baum welch starting for 2 Gaussian(s), iteration: 3 (1 of 1)
0%
Normalization for iteration: 3
Current Overall Likelihood Per Frame = 23.2946200241255
Convergence Ratio = 2.77501809408925
Baum welch starting for 2 Gaussian(s), iteration: 4 (1 of 1)
0%
Normalization for iteration: 4
Current Overall Likelihood Per Frame = 25.648902291918
Convergence Ratio = 2.35428226779247
Baum welch starting for 2 Gaussian(s), iteration: 5 (1 of 1)
0%
Normalization for iteration: 5
Current Overall Likelihood Per Frame = 26.339312424608
Convergence Ratio = 0.690410132689962
Baum welch starting for 2 Gaussian(s), iteration: 6 (1 of 1)
0%
Normalization for iteration: 6
Current Overall Likelihood Per Frame = 26.4127744270205
Split Gaussians, increase by 2
Current Overall Likelihood Per Frame = 26.4127744270205
Convergence Ratio = 0.073462002412505
Baum welch starting for 4 Gaussian(s), iteration: 1 (1 of 1)
0%
Normalization for iteration: 1
This step had 1635 ERROR messages and 0 WARNING messages. Please check the log file for details.
Current Overall Likelihood Per Frame = 25.9812545235223
Baum welch starting for 4 Gaussian(s), iteration: 2 (1 of 1)
0%
Normalization for iteration: 2
Current Overall Likelihood Per Frame = 29.3810373944511
Convergence Ratio = 3.39978287092885
Baum welch starting for 4 Gaussian(s), iteration: 3 (1 of 1)
0%
Normalization for iteration: 3
Current Overall Likelihood Per Frame = 33.7967671893848
Convergence Ratio = 4.4157297949337
Baum welch starting for 4 Gaussian(s), iteration: 4 (1 of 1)
0%
Normalization for iteration: 4
Current Overall Likelihood Per Frame = 39.2286248492159
Convergence Ratio = 5.43185765983112
Baum welch starting for 4 Gaussian(s), iteration: 5 (1 of 1)
0%
Normalization for iteration: 5
Current Overall Likelihood Per Frame = 39.9667068757539
Convergence Ratio = 0.738082026538024
Baum welch starting for 4 Gaussian(s), iteration: 6 (1 of 1)
0%
Normalization for iteration: 6
Current Overall Likelihood Per Frame = 40.0017852834741
Split Gaussians, increase by 4
Current Overall Likelihood Per Frame = 40.0017852834741
Convergence Ratio = 0.0350784077201709
Baum welch starting for 8 Gaussian(s), iteration: 1 (1 of 1)
0%
Normalization for iteration: 1
This step had 6941 ERROR messages and 0 WARNING messages. Please check the log file for details.
Current Overall Likelihood Per Frame = 39.5515681544029
Baum welch starting for 8 Gaussian(s), iteration: 2 (1 of 1)
0%
Normalization for iteration: 2
Current Overall Likelihood Per Frame = 43.9104342581423
Convergence Ratio = 4.35886610373944
Baum welch starting for 8 Gaussian(s), iteration: 3 (1 of 1)
0%
Normalization for iteration: 3
Current Overall Likelihood Per Frame = 50.497189384801
Convergence Ratio = 6.58675512665866
Baum welch starting for 8 Gaussian(s), iteration: 4 (1 of 1)
0%
Normalization for iteration: 4
Current Overall Likelihood Per Frame = 62.2500844390832
Convergence Ratio = 11.7528950542822
Baum welch starting for 8 Gaussian(s), iteration: 5 (1 of 1)
0%
Normalization for iteration: 5
Current Overall Likelihood Per Frame = 63.5334861278649
Convergence Ratio = 1.28340168878169
Baum welch starting for 8 Gaussian(s), iteration: 6 (1 of 1)
0%
Normalization for iteration: 6
Current Overall Likelihood Per Frame = 63.6126658624849
Split Gaussians, increase by 0
Training for 8 Gaussian(s) completed after 6 iterations
MODULE: 60 Lattice Generation
Skipped: $ST::CFG_MMIE set to 'no' in sphinx_train.cfg
MODULE: 61 Lattice Pruning
Skipped: $ST::CFG_MMIE set to 'no' in sphinx_train.cfg
MODULE: 62 Lattice Format Conversion
Skipped: $ST::CFG_MMIE set to 'no' in sphinx_train.cfg
MODULE: 65 MMIE Training
Skipped: $ST::CFG_MMIE set to 'no' in sphinx_train.cfg
MODULE: 90 deleted interpolation
Skipped for continuous models
这个问题纠结了很久，从0%可以看出，训练根本就没有启动，也查看了SPHINX的官方指导：http://cmusphinx.sourceforge.net/wiki/tutorialam。仍然不能的到解决。不过最终发现了问题所在：训练数据不足。官方虽然明确“1 hour of recording for command and control for single speaker”，但是我看“武林秘籍”作者既然可以训练出来，那么这里数据太少应该不适问题，故没有重视。但是经过N天的折腾之后，依然毫无改善，就去尝试增加训练数据，看看结果，没想到竟然真的是这里的问题。
我的训练数据离1hour还很远，只有“Estimated Total Hours Training: 0.0633277777777778”，但是相对于每个命令只录制一遍，有了相当大的改善（至少声学模型训练可以成功，而不是满天的ERROR）。之后将尝试继续添加训练数据，看看效果能否继续改善。
对于：Warning: Could not find Mic element这个问题，由于我是使用的Ubuntu的Wubi版，不知道和这个有没有关系，直接采取忽略的做法，原因当然是我说话，它能：
READY....
Listening...
Recording is stopped, start recording with ad_start_rec。
最终结果还不是很理想，“前进”偶尔会识别为“停止”，“右转”基本都识别为“停止”，估计原因是录制的音频不是很自然。之后将尝试添加更多训练数据以期改善。不过毕竟有了结果，这还是令人欣慰的。
运行结果如下：
$:pocketsphinx_continuous -hmm my_db.cd_cont_100 -lm my_db.lm.DMP -dict my_db.dic
INFO: cmd_ln.c(691): Parsing command line:
pocketsphinx_continuous \
-hmm my_db.cd_cont_100 \
-lm my_db.lm.DMP \
-dict my_db.dic
Current configuration:
[NAME] [DEFLT] [VALUE]
-adcdev
-agc none none
-agcthresh 2.0 2.000000e+00
-alpha 0.97 9.700000e-01
-argfile
-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
-bghist no no
-ceplen 13 13
-cmn current current
-cmninit 8.0 8.0
-compallsen no no
-debug 0
-dict my_db.dic
-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 my_db.cd_cont_100
-infile
-input_endian little little
-jsgf
-kdmaxbbi -1 -1
-kdmaxdepth 0 0
-kdtree
-latsize 5000 5000
-lda
-ldadim 0 0
-lextreedump 0 0
-lifter 0 0
-lm my_db.lm.DMP
-lmctl
-lmname default default
-logbase 1.0001 1.000100e+00
-logfn
-logspec no no
-lowerf 133.33334 1.333333e+02
-lpbeam 1e-40 1.000000e-40
-lponlybeam 7e-29 7.000000e-29
-lw 6.5 6.500000e+00
-maxhmmpf -1 -1
-maxnewoov 20 20
-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 40
-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-5 1.000000e-05
-pl_window 0 0
-rawlogdir
-remove_dc no no
-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
-time no no
-tmat
-tmatfloor 0.0001 1.000000e-04
-topn 4 4
-topn_beam 0 0
-toprule
-transform legacy legacy
-unit_area yes yes
-upperf 6855.4976 6.855498e+03
-usewdphones no no
-uw 1.0 1.000000e+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: cmd_ln.c(691): Parsing command line:
\
-alpha 0.97 \
-doublebw no \
-nfilt 40 \
-ncep 13 \
-lowerf 133.33334 \
-upperf 6855.4976 \
-nfft 512 \
-wlen 0.0256 \
-transform legacy \
-feat 1s_c_d_dd \
-agc none \
-cmn current \
-varnorm no
Current configuration:
[NAME] [DEFLT] [VALUE]
-agc none none
-agcthresh 2.0 2.000000e+00
-alpha 0.97 9.700000e-01
-ceplen 13 13
-cmn current current
-cmninit 8.0 8.0
-dither no no
-doublebw no no
-feat 1s_c_d_dd 1s_c_d_dd
-frate 100 100
-input_endian little little
-lda
-ldadim 0 0
-lifter 0 0
-logspec no no
-lowerf 133.33334 1.333333e+02
-ncep 13 13
-nfft 512 512
-nfilt 40 40
-remove_dc no no
-round_filters yes yes
-samprate 16000 1.600000e+04
-seed -1 -1
-smoothspec no no
-svspec
-transform legacy legacy
-unit_area yes yes
-upperf 6855.4976 6.855498e+03
-varnorm no no
-verbose no no
-warp_params
-warp_type inverse_linear inverse_linear
-wlen 0.025625 2.560000e-02
INFO: acmod.c(242): Parsed model-specific feature parameters from my_db.cd_cont_100/feat.params
INFO: feat.c(684): Initializing feature stream to type: '1s_c_d_dd', ceplen=13, CMN='current', VARNORM='no', AGC='none'
INFO: cmn.c(142): mean[0]= 12.00, mean[1..12]= 0.0
INFO: mdef.c(520): Reading model definition: my_db.cd_cont_100/mdef
INFO: bin_mdef.c(173): Allocating 166 * 8 bytes (1 KiB) for CD tree
INFO: tmat.c(205): Reading HMM transition probability matrices: my_db.cd_cont_100/transition_matrices
INFO: acmod.c(117): Attempting to use SCHMM computation module
INFO: ms_gauden.c(198): Reading mixture gaussian parameter: my_db.cd_cont_100/means
INFO: ms_gauden.c(292): 105 codebook, 1 feature, size:
INFO: ms_gauden.c(294): 8x39
INFO: ms_gauden.c(198): Reading mixture gaussian parameter: my_db.cd_cont_100/variances
INFO: ms_gauden.c(292): 105 codebook, 1 feature, size:
INFO: ms_gauden.c(294): 8x39
INFO: ms_gauden.c(354): 412 variance values floored
INFO: acmod.c(119): Attempting to use PTHMM computation module
INFO: ms_gauden.c(198): Reading mixture gaussian parameter: my_db.cd_cont_100/means
INFO: ms_gauden.c(292): 105 codebook, 1 feature, size:
INFO: ms_gauden.c(294): 8x39
INFO: ms_gauden.c(198): Reading mixture gaussian parameter: my_db.cd_cont_100/variances
INFO: ms_gauden.c(292): 105 codebook, 1 feature, size:
INFO: ms_gauden.c(294): 8x39
INFO: ms_gauden.c(354): 412 variance values floored
INFO: ptm_mgau.c(804): Number of codebooks doesn't match number of ciphones, doesn't look like PTM: 105 16
INFO: acmod.c(121): Falling back to general multi-stream GMM computation
INFO: ms_gauden.c(198): Reading mixture gaussian parameter: my_db.cd_cont_100/means
INFO: ms_gauden.c(292): 105 codebook, 1 feature, size:
INFO: ms_gauden.c(294): 8x39
INFO: ms_gauden.c(198): Reading mixture gaussian parameter: my_db.cd_cont_100/variances
INFO: ms_gauden.c(292): 105 codebook, 1 feature, size:
INFO: ms_gauden.c(294): 8x39
INFO: ms_gauden.c(354): 412 variance values floored
INFO: ms_senone.c(160): Reading senone mixture weights: my_db.cd_cont_100/mixture_weights
INFO: ms_senone.c(211): Truncating senone logs3(pdf) values by 10 bits
INFO: ms_senone.c(218): Not transposing mixture weights in memory
INFO: ms_senone.c(277): Read mixture weights for 105 senones: 1 features x 8 codewords
INFO: ms_senone.c(331): Mapping senones to individual codebooks
INFO: ms_mgau.c(122): The value of topn: 4
INFO: dict.c(306): Allocating 4104 * 20 bytes (80 KiB) for word entries
INFO: dict.c(321): Reading main dictionary: my_db.dic
INFO: dict.c(212): Allocated 0 KiB for strings, 0 KiB for phones
INFO: dict.c(324): 5 words read
INFO: dict.c(330): Reading filler dictionary: my_db.cd_cont_100/noisedict
INFO: dict.c(212): Allocated 0 KiB for strings, 0 KiB for phones
INFO: dict.c(333): 3 words read
INFO: dict2pid.c(396): Building PID tables for dictionary
INFO: dict2pid.c(404): Allocating 16^3 * 2 bytes (8 KiB) for word-initial triphones
INFO: dict2pid.c(131): Allocated 3136 bytes (3 KiB) for word-final triphones
INFO: dict2pid.c(195): Allocated 3136 bytes (3 KiB) for single-phone word triphones
INFO: ngram_model_arpa.c(77): No \data\ mark in LM file
INFO: ngram_model_dmp.c(142): Will use memory-mapped I/O for LM file
INFO: ngram_model_dmp.c(196): ngrams 1=7, 2=10, 3=13
INFO: ngram_model_dmp.c(242): 7 = LM.unigrams(+trailer) read
INFO: ngram_model_dmp.c(291): 10 = LM.bigrams(+trailer) read
INFO: ngram_model_dmp.c(317): 13 = LM.trigrams read
INFO: ngram_model_dmp.c(342): 4 = LM.prob2 entries read
INFO: ngram_model_dmp.c(362): 5 = LM.bo_wt2 entries read
INFO: ngram_model_dmp.c(382): 3 = LM.prob3 entries read
INFO: ngram_model_dmp.c(410): 1 = LM.tseg_base entries read
INFO: ngram_model_dmp.c(466): 7 = ascii word strings read
INFO: ngram_search_fwdtree.c(99): 5 unique initial diphones
INFO: ngram_search_fwdtree.c(147): 0 root, 0 non-root channels, 4 single-phone words
INFO: ngram_search_fwdtree.c(186): Creating search tree
INFO: ngram_search_fwdtree.c(191): before: 0 root, 0 non-root channels, 4 single-phone words
INFO: ngram_search_fwdtree.c(326): after: max nonroot chan increased to 138
INFO: ngram_search_fwdtree.c(338): after: 5 root, 10 non-root channels, 3 single-phone words
INFO: ngram_search_fwdflat.c(156): fwdflat: min_ef_width = 4, max_sf_win = 25
INFO: continuous.c(367): pocketsphinx_continuous COMPILED ON: Jul 29 2012, AT: 18:16:08
Warning: Could not find Mic element
READY....
Listening...
Recording is stopped, start recording with ad_start_rec
Stopped listening, please wait...
INFO: cmn_prior.c(121): 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(139): cmn_prior_update: to < 9.71 1.10 -0.35 -0.07 -0.45 -0.21 -0.27 0.05 -0.21 -0.26 -0.14 -0.11 -0.06 >
INFO: ngram_search_fwdtree.c(1549): 88 words recognized (1/fr)
INFO: ngram_search_fwdtree.c(1551): 1416 senones evaluated (16/fr)
INFO: ngram_search_fwdtree.c(1553): 548 channels searched (6/fr), 203 1st, 171 last
INFO: ngram_search_fwdtree.c(1557): 171 words for which last channels evaluated (1/fr)
INFO: ngram_search_fwdtree.c(1560): 38 candidate words for entering last phone (0/fr)
INFO: ngram_search_fwdtree.c(1562): fwdtree 0.01 CPU 0.014 xRT
INFO: ngram_search_fwdtree.c(1565): fwdtree 1.40 wall 1.632 xRT
INFO: ngram_search_fwdflat.c(305): Utterance vocabulary contains 3 words
INFO: ngram_search_fwdflat.c(940): 68 words recognized (1/fr)
INFO: ngram_search_fwdflat.c(942): 912 senones evaluated (11/fr)
INFO: ngram_search_fwdflat.c(944): 384 channels searched (4/fr)
INFO: ngram_search_fwdflat.c(946): 209 words searched (2/fr)
INFO: ngram_search_fwdflat.c(948): 101 word transitions (1/fr)
INFO: ngram_search_fwdflat.c(951): fwdflat -0.00 CPU -0.000 xRT
INFO: ngram_search_fwdflat.c(954): fwdflat 0.00 wall 0.003 xRT
INFO: ngram_search.c(1253): lattice start node <s>.0 end node </s>.83
INFO: ngram_search.c(1281): Eliminated 0 nodes before end node
INFO: ngram_search.c(1386): Lattice has 10 nodes, 8 links
INFO: ps_lattice.c(1352): Normalizer P(O) = alpha(</s>:83:84) = -258409
INFO: ps_lattice.c(1390): Joint P(O,S) = -258409 P(S|O) = 0
INFO: ngram_search.c(875): bestpath 0.00 CPU 0.000 xRT
INFO: ngram_search.c(878): bestpath 0.00 wall 0.000 xRT
000000000: 停止
READY....
Listening...
Recording is stopped, start recording with ad_start_rec
Stopped listening, please wait...
INFO: cmn_prior.c(121): cmn_prior_update: from < 9.71 1.10 -0.35 -0.07 -0.45 -0.21 -0.27 0.05 -0.21 -0.26 -0.14 -0.11 -0.06 >
INFO: cmn_prior.c(139): cmn_prior_update: to < 9.75 1.06 -0.31 -0.03 -0.41 -0.20 -0.24 0.06 -0.20 -0.28 -0.13 -0.14 -0.08 >
INFO: ngram_search_fwdtree.c(1549): 154 words recognized (2/fr)
INFO: ngram_search_fwdtree.c(1551): 1848 senones evaluated (23/fr)
INFO: ngram_search_fwdtree.c(1553): 718 channels searched (8/fr), 309 1st, 193 last
INFO: ngram_search_fwdtree.c(1557): 193 words for which last channels evaluated (2/fr)
INFO: ngram_search_fwdtree.c(1560): 8 candidate words for entering last phone (0/fr)
INFO: ngram_search_fwdtree.c(1562): fwdtree 0.02 CPU 0.020 xRT
INFO: ngram_search_fwdtree.c(1565): fwdtree 2.88 wall 3.601 xRT
INFO: ngram_search_fwdflat.c(305): Utterance vocabulary contains 3 words
INFO: ngram_search_fwdflat.c(940): 87 words recognized (1/fr)
INFO: ngram_search_fwdflat.c(942): 933 senones evaluated (12/fr)
INFO: ngram_search_fwdflat.c(944): 427 channels searched (5/fr)
INFO: ngram_search_fwdflat.c(946): 256 words searched (3/fr)
INFO: ngram_search_fwdflat.c(948): 120 word transitions (1/fr)
INFO: ngram_search_fwdflat.c(951): fwdflat 0.00 CPU 0.000 xRT
INFO: ngram_search_fwdflat.c(954): fwdflat 0.00 wall 0.003 xRT
INFO: ngram_search.c(1253): lattice start node <s>.0 end node </s>.71
INFO: ngram_search.c(1281): Eliminated 0 nodes before end node
INFO: ngram_search.c(1386): Lattice has 12 nodes, 9 links
INFO: ps_lattice.c(1352): Normalizer P(O) = alpha(</s>:71:78) = -227997
INFO: ps_lattice.c(1390): Joint P(O,S) = -227997 P(S|O) = 0
INFO: ngram_search.c(875): bestpath -0.00 CPU -0.000 xRT
INFO: ngram_search.c(878): bestpath 0.00 wall 0.000 xRT
000000001: 停止
READY....
Listening...
Recording is stopped, start recording with ad_start_rec
Stopped listening, please wait...
INFO: cmn_prior.c(121): cmn_prior_update: from < 9.75 1.06 -0.31 -0.03 -0.41 -0.20 -0.24 0.06 -0.20 -0.28 -0.13 -0.14 -0.08 >
INFO: cmn_prior.c(139): cmn_prior_update: to < 9.84 1.08 -0.32 -0.03 -0.43 -0.22 -0.24 0.04 -0.20 -0.27 -0.15 -0.13 -0.09 >
INFO: ngram_search_fwdtree.c(1549): 137 words recognized (1/fr)
INFO: ngram_search_fwdtree.c(1551): 2142 senones evaluated (23/fr)
INFO: ngram_search_fwdtree.c(1553): 801 channels searched (8/fr), 313 1st, 228 last
INFO: ngram_search_fwdtree.c(1557): 228 words for which last channels evaluated (2/fr)
INFO: ngram_search_fwdtree.c(1560): 56 candidate words for entering last phone (0/fr)
INFO: ngram_search_fwdtree.c(1562): fwdtree 0.02 CPU 0.017 xRT
INFO: ngram_search_fwdtree.c(1565): fwdtree 1.38 wall 1.487 xRT
INFO: ngram_search_fwdflat.c(305): Utterance vocabulary contains 4 words
INFO: ngram_search_fwdflat.c(940): 108 words recognized (1/fr)
INFO: ngram_search_fwdflat.c(942): 1455 senones evaluated (16/fr)
INFO: ngram_search_fwdflat.c(944): 590 channels searched (6/fr)
INFO: ngram_search_fwdflat.c(946): 351 words searched (3/fr)
INFO: ngram_search_fwdflat.c(948): 154 word transitions (1/fr)
INFO: ngram_search_fwdflat.c(951): fwdflat 0.00 CPU 0.000 xRT
INFO: ngram_search_fwdflat.c(954): fwdflat 0.00 wall 0.004 xRT
INFO: ngram_search.c(1253): lattice start node <s>.0 end node </s>.90
INFO: ngram_search.c(1281): Eliminated 0 nodes before end node
INFO: ngram_search.c(1386): Lattice has 13 nodes, 12 links
INFO: ps_lattice.c(1352): Normalizer P(O) = alpha(</s>:90:91) = -208439
INFO: ps_lattice.c(1390): Joint P(O,S) = -208439 P(S|O) = 0
INFO: ngram_search.c(875): bestpath -0.00 CPU -0.000 xRT
INFO: ngram_search.c(878): bestpath 0.00 wall 0.000 xRT
000000002: 左转
READY....
内容来自用户分享和网络整理，不保证内容的准确性，如有侵权内容，可联系管理员处理 
相关文章推荐
- Pocketsphinx语音识别--重新训练声学模型
- 【sphinx]声学模型训练流程学习
- 语音识别工具包pocketsphinx-0.8声学模型训练
- CMUSphinx适应声学模型
- sphinx中自己提取特征参数训练声学模型参数方法探讨
- 【sphinx】中文声学模型训练
- PocketSphinx语音识别系统----声学模型的训练与使用
- 语音识别工具包pocketsphinx-0.8声学模型训练
- 机器学习中的数据不平衡问题----通过随机采样比例大的类别使得训练集中大类的个数与小类相当，或者模型中加入惩罚项
- Caffe训练AlexNet网络模型——问题三
- Keras 循环训练模型跑数据时内存泄漏的问题解决办法
- 训练深度模型的优化问题（七）
- 训练深度模型的优化问题（十八）
- 基于拉格尔函数模型的预测控制在线自辩识问题
- Kaldi 训练一个 DNN 声学模型
- windows下训练FCN模型所遇到的问题
- python训练模型上线问题总结
- windows下训练FCN模型所遇到的问题
- 用训练好的模型测试单张图片存在的问题