-
Notifications
You must be signed in to change notification settings - Fork 56
/
Copy pathMakefile
553 lines (427 loc) · 21.7 KB
/
Makefile
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
SHELL := /bin/bash
# Use this file to override various settings
-include Makefile.options
DO_LANGUAGE_DETECTION?=yes
# Set to 'yes' if you want to do speaker ID for trs files
# Assumes you have models for speaker ID
DO_SPEAKER_ID?=yes
SID_SIMILARITY_THRESHOLD?=13
SPEAKER_ID_SERVER_URL?=''
# Where is Kaldi root directory?
KALDI_ROOT?=/home/speech/tools/kaldi-trunk
# Location of the Java binary
JAVA_BIN?=/usr/bin/java
# How many processes to use for one transcription task
njobs ?= 1
# How many threads to use in each process
nthreads ?= 1
PATH:=utils:$(KALDI_ROOT)/src/bin:$(KALDI_ROOT)/tools/openfst/bin:$(KALDI_ROOT)/src/fstbin/:$(KALDI_ROOT)/src/gmmbin/:$(KALDI_ROOT)/src/featbin/:$(KALDI_ROOT)/src/lm/:$(KALDI_ROOT)/src/sgmmbin/:$(KALDI_ROOT)/src/sgmm2bin/:$(KALDI_ROOT)/src/fgmmbin/:$(KALDI_ROOT)/src/latbin/:$(KALDI_ROOT)/src/nnet2bin/:$(KALDI_ROOT)/src/online2bin/:$(KALDI_ROOT)/src/kwsbin:$(KALDI_ROOT)/src/lmbin:$(PATH):$(KALDI_ROOT)/src/ivectorbin:$(KALDI_ROOT)/src/nnet3bin:$(KALDI_ROOT)/src/rnnlmbin:$(PATH)
# Needed for compounder.py
LD_LIBRARY_PATH:=$(KALDI_ROOT)/tools/openfst/lib:$(LD_LIBRARY_PATH)
export train_cmd=run.pl
export decode_cmd=run.pl
export cuda_cmd=run.pl
export mkgraph_cmd=run.pl
# Main language model (should be slightly pruned), used for rescoring
#LM ?=language_model/interpolated.pruned9.4g.arpa.gz
# More aggressively pruned LM, used in decoding
PRUNED_LM ?=language_model/interpolated.pruned9.4g.arpa.gz
RNNLM_MODEL ?=language_model/rnnlm
#COMPOUNDER_LM ?=language_model/compounder-pruned.vestlused-dev.splitw.arpa.gz
COMPOUNDER_LM ?=language_model/compounder.pruned9.4g.arpa.gz
ACOUSTIC_MODEL?=cnn_tdnn_1d_online
# Vocabulary in dict format (no pronouncation probs for now)
VOCAB?=language_model/lexicon.txt
ET_G2P_FST?=../et-g2p-fst
LM_SCALE?=10
DO_PUNCTUATION?=no
ifeq "yes" "$(DO_PUNCTUATION)"
PUNCTUATE_JSON_CMD?=cat
DOT_PUNCTUATED=.punctuated
endif
# Find out where this Makefile is located (this is not really needed)
where-am-i = $(lastword $(MAKEFILE_LIST))
THIS_DIR := $(shell dirname $(call where-am-i))
FINAL_PASS=$(ACOUSTIC_MODEL)_pruned_rnnlm_unk
LD_LIBRARY_PATH:=$(KALDI_ROOT)/tools/openfst/lib:$(LD_LIBRARY_PATH)
.SECONDARY:
.DELETE_ON_ERROR:
PYTHONIOENCODING="utf-8"
export
# Call this (once) before using the system
.init: .kaldi .lang
.kaldi:
rm -f steps utils sid rnnlm
ln -s $(KALDI_ROOT)/egs/wsj/s5/steps
ln -s $(KALDI_ROOT)/egs/wsj/s5/utils
ln -s $(KALDI_ROOT)/egs/sre08/v1/sid
ln -s $(KALDI_ROOT)/scripts/rnnlm
mkdir -p src-audio
#.lang: build/fst/data/prunedlm_unk build/fst/$(ACOUSTIC_MODEL)/graph_prunedlm_unk build/fst/data/largelm_unk build/fst/data/rnnlm_unk build/fst/data/compounderlm
.lang: build/fst/data/prunedlm_unk build/fst/$(ACOUSTIC_MODEL)/graph_prunedlm_unk build/fst/data/rnnlm_unk build/fst/data/compounderlm
build/fst/$(ACOUSTIC_MODEL)/final.mdl:
rm -rf `dirname $@`
mkdir -p `dirname $@`
cp -r $(THIS_DIR)/kaldi-data/$(ACOUSTIC_MODEL)/* `dirname $@`
perl -i -npe 's#=.*online/#=build/fst/$(ACOUSTIC_MODEL)/#' build/fst/$(ACOUSTIC_MODEL)/conf/*.conf
if [ ! -e build/fst/$(ACOUSTIC_MODEL)/cmvn_opts ]; then \
echo "--norm-means=false --norm-vars=false" > build/fst/$(ACOUSTIC_MODEL)/cmvn_opts; \
fi
build/fst/data/dict/.done: $(VOCAB) build/fst/$(ACOUSTIC_MODEL)/final.mdl
rm -rf build/fst/data/dict
mkdir -p build/fst/data/dict
cp -r $(THIS_DIR)/kaldi-data/dict/* build/fst/data/dict
rm -f build/fst/data/dict/lexicon.txt build/fst/data/dict/lexiconp.txt
cat models/etc/filler16k.dict | egrep -v "^<.?s>" > build/fst/data/dict/lexicon.txt
cat $(VOCAB) | perl -npe 's/\(\d\)(\s)/\1/' >> build/fst/data/dict/lexicon.txt
touch -m $@
build/fst/data/prunedlm: $(PRUNED_LM) $(VOCAB) build/fst/$(ACOUSTIC_MODEL)/final.mdl build/fst/data/dict/.done
rm -rf build/fst/data/prunedlm
mkdir -p build/fst/data/prunedlm
utils/prepare_lang.sh --phone-symbol-table build/fst/$(ACOUSTIC_MODEL)/phones.txt build/fst/data/dict '<unk>' build/fst/data/dict/tmp build/fst/data/prunedlm
gunzip -c $(PRUNED_LM) | arpa2fst --disambig-symbol=#0 \
--read-symbol-table=build/fst/data/prunedlm/words.txt - build/fst/data/prunedlm/G.fst
echo "Checking how stochastic G is (the first of these numbers should be small):"
fstisstochastic build/fst/data/prunedlm/G.fst || echo "not stochastic (probably OK)"
utils/validate_lang.pl build/fst/data/prunedlm || exit 1
build/fst/data/unk_lang_model: build/fst/data/dict/.done
rm -rf $@
utils/lang/make_unk_lm.sh build/fst/data/dict $@
build/fst/data/prunedlm_unk: build/fst/data/unk_lang_model build/fst/data/prunedlm
rm -rf $@
utils/prepare_lang.sh --unk-fst build/fst/data/unk_lang_model/unk_fst.txt build/fst/data/dict "<unk>" build/fst/data/prunedlm $@
cp build/fst/data/prunedlm/G.fst $@
build/fst/%/graph_prunedlm_unk: build/fst/data/prunedlm_unk build/fst/%/final.mdl
rm -rf $@
self_loop_scale_arg=""; \
if [ -f build/fst/$*/frame_subsampling_factor ]; then \
factor=`cat build/fst/$*/frame_subsampling_factor`; \
if [ $$factor -eq "3" ]; then \
self_loop_scale_arg="--self-loop-scale 1.0 "; \
fi; \
fi; \
utils/mkgraph.sh $$self_loop_scale_arg build/fst/data/prunedlm_unk build/fst/$* $@
rm -rf build/fst/data/prunedlm_unk/tmp
touch -m $@
build/fst/data/largelm_unk: build/fst/data/prunedlm
rm -rf $@
mkdir -p $@
utils/build_const_arpa_lm.sh \
$(LM) build/fst/data/prunedlm $@
build/fst/data/rnnlm_unk: $(RNNLM_MODEL) build/fst/data/prunedlm
rm -rf $@
mkdir -p $@
cp -r $(RNNLM_MODEL)/* $@/
cp build/fst/data/prunedlm/words.txt $@/config/words.txt
brk_id=`cat $@/config/words.txt | wc -l`; \
echo "<brk> $$brk_id" >> $@/config/words.txt; \
bos_id=`grep "^<s>" $@/config/words.txt | awk '{print $$2}'`; \
eos_id=`grep "^</s>" $@/config/words.txt | awk '{print $$2}'`; \
echo "--eos-symbol=$${eos_id} --brk-symbol=$${brk_id} --bos-symbol=$${bos_id}" > $@/special_symbol_opts.txt
rnnlm/get_word_features.py \
--unigram-probs $@/config/unigram_probs.txt \
build/fst/data/prunedlm/words.txt \
$@/config/features.txt \
> $@/word_feats.txt
build/fst/data/compounderlm: $(COMPOUNDER_LM) $(VOCAB)
rm -rf $@
mkdir -p $@
cat $(VOCAB) | perl -npe 's/(\(\d\))?\s.+//' | uniq | ./scripts/make-compounder-symbols.py > $@/words.txt
zcat $(COMPOUNDER_LM) | \
grep -v '<s> <s>' | \
grep -v '</s> <s>' | \
grep -v '</s> </s>' | \
arpa2fst --disambig-symbol='#0' --read-symbol-table=$@/words.txt - | fstproject --project_output=true | fstarcsort --sort_type=ilabel > $@/G.fst
build/fst/%/graph_prunedlm: build/fst/data/prunedlm build/fst/%/final.mdl
rm -rf $@
utils/mkgraph.sh --self-loop-scale 1.0 build/fst/data/prunedlm build/fst/$* $@
build/audio/base/%.wav: src-audio/%.wav
mkdir -p `dirname $@`
sox $^ -c 1 -b 16 build/audio/base/$*.wav rate -v 16k
build/audio/base/%.wav: src-audio/%.mp3
mkdir -p `dirname $@`
ffmpeg -i $^ -f sox - | sox -t sox - -c 1 -b 16 $@ rate -v 16k
build/audio/base/%.wav: src-audio/%.ogg
mkdir -p `dirname $@`
sox $^ -c 1 build/audio/base/$*.wav rate -v 16k
build/audio/base/%.wav: src-audio/%.mp2
mkdir -p `dirname $@`
sox $^ -c 1 build/audio/base/$*.wav rate -v 16k
build/audio/base/%.wav: src-audio/%.m4a
mkdir -p `dirname $@`
ffmpeg -i $^ -f sox - | sox -t sox - -c 1 -b 16 $@ rate -v 16k
build/audio/base/%.wav: src-audio/%.mp4
mkdir -p `dirname $@`
ffmpeg -i $^ -f sox - | sox -t sox - -c 1 -b 16 $@ rate -v 16k
build/audio/base/%.wav: src-audio/%.flac
mkdir -p `dirname $@`
sox $^ -c 1 build/audio/base/$*.wav rate -v 16k
build/audio/base/%.wav: src-audio/%.amr
mkdir -p `dirname $@`
amrnb-decoder $^ [email protected]
sox -s -b 16 -c 1 -r 8000 [email protected] -c 1 build/audio/base/$*.wav rate -v 16k
build/audio/base/%.wav: src-audio/%.mpg
mkdir -p `dirname $@`
ffmpeg -i $^ -f sox - | sox -t sox - -c 1 -b 16 build/audio/base/$*.wav rate -v 16k
# Speaker diarization
build/diarization/%/show.seg: build/audio/base/%.wav
rm -rf `dirname $@`
mkdir -p `dirname $@`
python3 ./scripts/find_speech_segments.py $^ `dirname $@`/show.uem.seg; \
./scripts/diarization.sh $^ `dirname $@`/show.uem.seg
build/trans/%/test.pre_lid/wav.scp:
mkdir -p build/trans/$*/test.pre_lid
echo "$* build/audio/base/$*.wav" > $@
build/trans/%/test.pre_lid/reco2file_and_channel:
echo "$* $* A" > $@
# if diarization doesn't find andy speech segments,
# we generate a 'dummy' short speech segment,
# so that decoding won't fail
# this is unfortunately pretty ugly
build/trans/%/test.pre_lid/segments: build/diarization/%/show.seg build/trans/%/test.pre_lid/wav.scp build/trans/%/test.pre_lid/reco2file_and_channel
cat build/diarization/$*/show.seg | cut -f 3,4,8 -d " " | \
while read LINE ; do \
len_in_frames=`echo $$LINE | cut -f 2 -d " "`; \
start=`echo $$LINE | cut -f 1,2 -d " " | perl -ne '@t=split(); $$start=$$t[0]/100.0; printf("%08.3f", $$start);'`; \
end=`echo $$LINE | cut -f 1,2 -d " " | perl -ne '@t=split(); $$start=$$t[0]/100.0; $$len=$$t[1]/100.0; $$end=$$start+$$len; printf("%08.3f", $$end);'`; \
sp_id=`echo $$LINE | cut -f 3 -d " "`; \
if [ $${len_in_frames} -gt 30 ]; then \
echo $*-$${sp_id}---$${start}-$${end} $* $$start $$end; \
fi; \
done > $@
if [ ! -s $@ ]; then \
echo "$*-dummy---0.000-0.110 $* 0.0 0.110" > $@; \
fi
ifeq "yes" "$(DO_LANGUAGE_DETECTION)"
build/trans/%/segments: build/trans/%/test.pre_lid/segments build/trans/%/test.pre_lid/utt2lang
cp build/trans/$*/test.pre_lid/{wav.scp,reco2file_and_channel} build/trans/$*
# Remove segments whose language is not "et"
grep "et$$" build/trans/$*/test.pre_lid/utt2lang | sort | \
join <(sort build/trans/$*/test.pre_lid/segments) - | \
awk '{print($$1, $$2, $$3, $$4)}' | LC_ALL=C sort > $@
else
build/trans/%/segments: build/trans/%/test.pre_lid/segments
cp build/trans/$*/test.pre_lid/{wav.scp,reco2file_and_channel,segments} build/trans/$*
endif
build/trans/%/wav.scp: build/trans/%/test.pre_lid/wav.scp
cp $^ $@
build/trans/%/utt2spk: build/trans/%/segments
cat build/trans/$*/segments | \
perl -npe 's/\s+.*//; s/((.*)---.*)/\1 \2/' > $@
build/trans/%/spk2utt: build/trans/%/utt2spk
utils/utt2spk_to_spk2utt.pl $^ > $@
# MFCC calculation
build/trans/%/mfcc: build/trans/%/spk2utt build/trans/%/segments build/fst/$(ACOUSTIC_MODEL)/final.mdl
rm -rf $@
rm -f build/trans/$*/cmvn.scp build/trans/$*/mfcc.scp
steps/make_mfcc.sh --mfcc-config build/fst/$(ACOUSTIC_MODEL)/conf/mfcc.conf --cmd "$$decode_cmd" --nj $(njobs) \
build/trans/$* build/trans/$*/exp/make_mfcc $@ || exit 1
steps/compute_cmvn_stats.sh build/trans/$* build/trans/$*/exp/make_mfcc $@ || exit 1
utils/fix_data_dir.sh build/trans/$*
# Touch files that utils/fix_data_dir.sh might modify, in the right order
# so that make will not try to remake them
touch -m build/trans/$*/wav.scp
touch -m build/trans/$*/segments
touch -m build/trans/$*/utt2spk
touch -m build/trans/$*/spk2utt
touch -m $@
build/trans/%/ivectors: build/trans/%/mfcc
rm -rf $@
steps/online/nnet2/extract_ivectors_online.sh --cmd "$$decode_cmd" --nj $(njobs) \
build/trans/$* build/fst/$(ACOUSTIC_MODEL)/ivector_extractor $@ || exit 1;
### Do 1-pass decoding using chain online models
build/trans/%/$(ACOUSTIC_MODEL)_pruned_unk/decode/log: build/fst/$(ACOUSTIC_MODEL)/final.mdl build/fst/$(ACOUSTIC_MODEL)/graph_prunedlm_unk build/trans/%/spk2utt build/trans/%/mfcc build/trans/%/ivectors
rm -rf build/trans/$*/$(ACOUSTIC_MODEL)_pruned_unk
mkdir -p build/trans/$*/$(ACOUSTIC_MODEL)_pruned_unk
(cd build/trans/$*/$(ACOUSTIC_MODEL)_pruned_unk; for f in ../../../fst/$(ACOUSTIC_MODEL)/*; do ln -s $$f; done)
steps/nnet3/decode.sh --num-threads $(nthreads) --acwt 1.0 --post-decode-acwt 10.0 \
--skip-scoring true --cmd "$$decode_cmd" --nj $(njobs) \
--online-ivector-dir build/trans/$*/ivectors \
--skip-diagnostics true \
build/fst/$(ACOUSTIC_MODEL)/graph_prunedlm_unk build/trans/$* `dirname $@` || exit 1;
(cd build/trans/$*/$(ACOUSTIC_MODEL)_pruned_unk; ln -s ../../../fst/$(ACOUSTIC_MODEL)/graph_prunedlm_unk graph)
# Rescore lattices with a larger language model
build/trans/%/$(ACOUSTIC_MODEL)_pruned_rescored_main_unk/decode/log: build/trans/%/$(ACOUSTIC_MODEL)_pruned_unk/decode/log build/fst/data/largelm_unk
rm -rf build/trans/$*/$(ACOUSTIC_MODEL)_pruned_rescored_main_unk
mkdir -p build/trans/$*/$(ACOUSTIC_MODEL)_pruned_rescored_main_unk
(cd build/trans/$*/$(ACOUSTIC_MODEL)_pruned_rescored_main_unk; for f in ../../../fst/$(ACOUSTIC_MODEL)/*; do ln -s $$f; done)
steps/lmrescore_const_arpa.sh \
build/fst/data/prunedlm_unk build/fst/data/largelm_unk \
build/trans/$* \
build/trans/$*/$(ACOUSTIC_MODEL)_pruned_unk/decode build/trans/$*/$(ACOUSTIC_MODEL)_pruned_rescored_main_unk/decode || exit 1;
cp -r --preserve=links build/trans/$*/$(ACOUSTIC_MODEL)_pruned_unk/graph build/trans/$*/$(ACOUSTIC_MODEL)_pruned_rescored_main_unk/
build/trans/%/$(ACOUSTIC_MODEL)_pruned_rescored_main_rnnlm_unk/decode/log: build/trans/%/$(ACOUSTIC_MODEL)_pruned_rescored_main_unk/decode/log build/fst/data/rnnlm_unk
rm -rf build/trans/$*/$(ACOUSTIC_MODEL)_pruned_rescored_main_rnnlm_unk
mkdir -p build/trans/$*/$(ACOUSTIC_MODEL)_pruned_rescored_main_rnnlm_unk
(cd build/trans/$*/$(ACOUSTIC_MODEL)_pruned_rescored_main_rnnlm_unk; for f in ../../../fst/$(ACOUSTIC_MODEL)/*; do ln -s $$f; done)
rnnlm/lmrescore_pruned.sh \
--skip-scoring true \
--max-ngram-order 4 \
build/fst/data/largelm_unk \
build/fst/data/rnnlm_unk \
build/trans/$* \
build/trans/$*/$(ACOUSTIC_MODEL)_pruned_rescored_main_unk/decode \
build/trans/$*/$(ACOUSTIC_MODEL)_pruned_rescored_main_rnnlm_unk/decode
cp -r --preserve=links build/trans/$*/$(ACOUSTIC_MODEL)_pruned_unk/graph build/trans/$*/$(ACOUSTIC_MODEL)_pruned_rescored_main_rnnlm_unk/
build/trans/%/$(ACOUSTIC_MODEL)_pruned_rnnlm_unk/decode/log: build/trans/%/$(ACOUSTIC_MODEL)_pruned_unk/decode/log build/fst/data/rnnlm_unk
rm -rf build/trans/$*/$(ACOUSTIC_MODEL)_pruned_rnnlm_unk
mkdir -p build/trans/$*/$(ACOUSTIC_MODEL)_pruned_rnnlm_unk
(cd build/trans/$*/$(ACOUSTIC_MODEL)_pruned_rnnlm_unk; for f in ../../../fst/$(ACOUSTIC_MODEL)/*; do ln -s $$f; done)
rnnlm/lmrescore_pruned.sh \
--skip-scoring true \
--max-ngram-order 4 \
build/fst/data/prunedlm_unk \
build/fst/data/rnnlm_unk \
build/trans/$* \
build/trans/$*/$(ACOUSTIC_MODEL)_pruned_unk/decode \
build/trans/$*/$(ACOUSTIC_MODEL)_pruned_rnnlm_unk/decode
cp -r --preserve=links build/trans/$*/$(ACOUSTIC_MODEL)_pruned_unk/graph build/trans/$*/$(ACOUSTIC_MODEL)_pruned_rnnlm_unk/
%/decode/.ctm: %/decode/log
frame_shift_opt=""; \
if [ -f $*/frame_subsampling_factor ]; then \
factor=`cat $*/frame_subsampling_factor`; \
frame_shift_opt="--frame-shift 0.0$$factor"; \
fi; \
steps/get_ctm.sh $$frame_shift_opt `dirname $*` $*/graph $*/decode
touch -m $@
%_unk/decode/.ctm: %_unk/decode/log
frame_shift_opt=""; \
if [ -f $*_unk/frame_subsampling_factor ]; then \
factor=`cat $*_unk/frame_subsampling_factor`; \
frame_shift_opt="--frame-shift 0.0$$factor"; \
fi; \
$(THIS_DIR)/local/get_ctm_unk.sh --use_segments false $$frame_shift_opt \
--unk-p2g-cmd "python3 $(THIS_DIR)/local/unk_p2g.py --p2g-cmd 'python3 $(ET_G2P_FST)/g2p.py --inverse --fst $(ET_G2P_FST)/data/chars.fst --nbest 1'" \
--unk-word '<unk>' \
--min-lmwt $(LM_SCALE) \
--max-lmwt $(LM_SCALE) \
`dirname $*` $*_unk/graph $*_unk/decode
touch -m $@
build/trans/%.segmented.splitw2.ctm: build/trans/%/decode/.ctm
cat build/trans/$*/decode/score_$(LM_SCALE)/`dirname $*`.ctm | perl -npe 's/(.*)-(S\d+)---(\S+)/\1_\3_\2/' > $@
%.with-compounds.ctm: %.splitw2.ctm build/fst/data/compounderlm
python3 scripts/compound-ctm.py \
"python3 scripts/compounder.py build/fst/data/compounderlm/G.fst build/fst/data/compounderlm/words.txt" \
< $*.splitw2.ctm > $@
%.segmented.ctm: %.segmented.with-compounds.ctm
cat $^ | grep -v "++" | grep -v "\[sil\]" | grep -v -e " $$" | perl -npe 's/\+//g' | sort -k1,1 -k 3,3g > $@
ifeq "yes" "$(DO_SPEAKER_ID)"
build/trans/%/$(FINAL_PASS).json: build/trans/%/$(FINAL_PASS).segmented.ctm build/sid/%/sid-result.json build/diarization/%/show.seg
python3 local/segmented_ctm2json.py --speaker-names build/sid/$*/sid-result.json --pms-seg build/diarization/$*/show.uem.seg build/trans/$*/$(FINAL_PASS).segmented.ctm > $@
else
build/trans/%/$(FINAL_PASS).json: build/trans/%/$(FINAL_PASS).segmented.ctm build/diarization/%/show.seg
python3 local/segmented_ctm2json.py --pms-seg build/diarization/$*/show.uem.seg build/trans/$*/$(FINAL_PASS).segmented.ctm > $@
endif
%.with-compounds.synced.ctm: %.segmented.with-compounds.ctm
cat $^ | ./scripts/unsegment-ctm.py | LC_ALL=C sort -k 1,1 -k 3,3n -k 4,4n > $@
%.synced.ctm: %.segmented.ctm
cat $^ | ./scripts/unsegment-ctm.py | LC_ALL=C sort -k 1,1 -k 3,3n -k 4,4n > $@
%.ctm: %.synced.ctm
cat $^ | grep -v "<" > $@
%.with-sil.ctm: %.ctm
cat $^ | ./scripts/ctm2with-sil-ctm.py > $@
%.punctuated.json: %.json
cat $^ | $(PUNCTUATE_JSON_CMD) > $@
%.normalized.json: %.json
./local/normalize_json.py ./local/words2numbers.py $^ > $@
%.hyp: %.segmented.ctm
cat $^ | ./scripts/segmented-ctm-to-hyp.py > $@
build/trans/%/$(FINAL_PASS)$(DOT_PUNCTUATED).trs: build/trans/%/$(FINAL_PASS)$(DOT_PUNCTUATED).normalized.json
./local/json2trs.py --fid $* $^ > $@
%.srt: %.json
./local/json2srt.py $^ > $@
%.txt: %.trs
cat $^ | grep -v "^<" > $@
build/output/%.json: build/trans/%/$(FINAL_PASS)$(DOT_PUNCTUATED).normalized.json
mkdir -p `dirname $@`
cp $^ $@
build/output/%.trs: build/trans/%/$(FINAL_PASS)$(DOT_PUNCTUATED).trs
mkdir -p `dirname $@`
cp $^ $@
build/output/%.ctm: build/trans/%/$(FINAL_PASS).ctm
mkdir -p `dirname $@`
cp $^ $@
build/output/%.txt: build/trans/%/$(FINAL_PASS)$(DOT_PUNCTUATED).normalized.txt
mkdir -p `dirname $@`
cp $^ $@
build/output/%.with-compounds.ctm: build/trans/%/$(FINAL_PASS).with-compounds.ctm
mkdir -p `dirname $@`
cp $^ $@
build/output/%.srt: build/trans/%/$(FINAL_PASS)$(DOT_PUNCTUATED).normalized.srt
mkdir -p `dirname $@`
cp $^ $@
### Speaker ID stuff
ifeq ($(SPEAKER_ID_SERVER_URL), '')
# MFCC for Speaker ID, since the features for MFCC are different from speech recognition
build/sid/%/wav.scp: build/trans/%/wav.scp
mkdir -p `dirname $@`
rm -f $@
ln $^ $@
build/sid/%/utt2spk : build/trans/%/utt2spk
mkdir -p `dirname $@`
rm -f $@
ln $^ $@
build/sid/%/spk2utt : build/trans/%/spk2utt
mkdir -p `dirname $@`
rm -f $@
ln $^ $@
build/sid/%/segments : build/trans/%/segments
mkdir -p `dirname $@`
rm -f $@
ln $^ $@
build/sid/%/mfcc: build/sid/%/wav.scp build/sid/%/utt2spk build/sid/%/spk2utt build/sid/%/segments
rm -rf $@
rm -f build/sid/$*/vad.scp
rm -f build/sid/$*/cmvn.scp
steps/make_mfcc.sh --mfcc-config conf/mfcc_sid.conf --cmd "$$train_cmd" --nj $(njobs) \
build/sid/$* build/sid/$*/exp/make_mfcc $@ || exit 1
steps/compute_cmvn_stats.sh build/sid/$* build/sid/$*/exp/make_mfcc $@ || exit 1
sid/compute_vad_decision.sh --nj $(njobs) --cmd "$$decode_cmd" \
build/sid/$* build/sid/$*/exp/make_vad $@ || exit 1
# i-vectors for each speaker in our audio file
build/sid/%/ivectors: build/sid/%/mfcc
rm -rf build/sid/$*/ivectors
sid/extract_ivectors.sh --cmd "$$decode_cmd" --nj $(njobs) --num-threads $(nthreads) \
$(THIS_DIR)/kaldi-data/sid/extractor_2048 build/sid/$* $@
# cross-product between trained speakers and diarized speakers
build/sid/%/trials: build/sid/%/ivectors
join -j 2 \
<(cut -d " " -f 1 kaldi-data/sid/name_ivector.scp | sort ) \
<(cut -d " " -f 1 build/sid/$*/ivectors/spk_ivector.scp | sort ) > $@
build/sid/%/lda_plda_scores: build/sid/%/trials
ivector-plda-scoring --normalize-length=true \
"ivector-copy-plda --smoothing=0.3 kaldi-data/sid/lda_plda - |" \
"ark:ivector-subtract-global-mean scp:kaldi-data/sid//name_ivector.scp ark:- | transform-vec kaldi-data/sid/transform.mat ark:- ark:- | ivector-normalize-length ark:- ark:- |" \
"ark:ivector-subtract-global-mean kaldi-data/sid/mean.vec scp:build/sid/$*/ivectors/spk_ivector.scp ark:- | transform-vec kaldi-data/sid/transform.mat ark:- ark:- | ivector-normalize-length ark:- ark:- |" \
build/sid/$*/trials $@
build/sid/%/sid-result.json: build/sid/%/lda_plda_scores
cat build/sid/$*/lda_plda_scores | sort -k2,2 -k3,3nr | awk '{print $$3, $$1, $$2}' | uniq -f2 | awk '{if ($$1 > $(SID_SIMILARITY_THRESHOLD)) {print $$3, $$2}}' | \
perl -npe 's/^\S+-(S\d+)/\1/; s/_/ /g;' | python -c 'import json, sys; spks={s.split()[0]:{"name" : " ".join(s.split()[1:])} for s in sys.stdin}; json.dump(spks, sys.stdout);' > $@
else
build/sid/%/wav_segments: build/trans/%/spk2utt build/trans/%/wav.scp
utils/data/extract_wav_segments_data_dir.sh --cmd "$$decode_cmd" --nj $(njobs) build/trans/$* $@
perl -i -npe 's/^\S+-(S\d+)/\1/;' build/sid/$*/wav_segments/spk2utt
build/sid/%/sid-result.json: build/sid/%/wav_segments
PATH=$(KALDI_ROOT)/src/featbin:$$PATH; ./local/speaker-id-from-server.py --url $(SPEAKER_ID_SERVER_URL) build/sid/$*/wav_segments/spk2utt build/sid/$*/wav_segments/wav.scp $@
endif
## Language ID
build/lid/%: build/trans/%/test.pre_lid/segments
rm -rf $@
mkdir -p $@
python3 local/extract_lid_features_kaldi.py build/trans/$*/test.pre_lid $@
cat build/trans/$*/test.pre_lid/segments | awk '{print($$1, "0")}' > build/lid/$*/trials
build/trans/%/test.pre_lid/utt2lang: build/lid/%
threshold=`cat models/lid_et/threshold`; \
ivector-subtract-global-mean models/lid_et/xvector.global.vec scp:build/lid/$*/xvector.scp ark:- | \
ivector-normalize-length --scaleup=false ark:- ark:- | \
logistic-regression-eval --apply-log=true --max-steps=20 --mix-up=0 \
models/lid_et/lr.scale.model \
ark:build/lid/$*/trials ark:- - | \
awk '{print($$1, $$3 > '$$threshold' ? "et" : "other")}' > $@
# Meta-target that deletes all files created during processing a file. Call e.g. 'make .etteytlus2013.clean
.%.clean:
rm -rf build/audio/base/$*.wav build/audio/segmented/$* build/diarization/$* build/trans/$* build/sid/$*
# Also deletes the output files
.%.cleanest: .%.clean
rm -rf build/output/$*.{trs,txt,ctm,with-compounds.ctm,srt,json}