api doc

# This file will be some notes on the api func in details.


# ark defaults to bytes,and 'ark ,t' refers to text

# Create MFCC feature files or stream.
# Usage:  compute-mfcc-feats [options...] <wav-rspecifier> <feats-wspecifier>
compute-mfcc-feats --verbose=2 --config=$mfcc.conf "scp:echo utterance-id1 /home/zhang/Projects/AI2/VoiceAppFile/VoiceAns/2/0311203006_PAPER-934712-QT-742440_QT-742440ans.wav|" ark[,t]:$mfcc.feat.file (or -)




# $1 is the mfcc-features-file or stream('-'), '-' means output data stream in the shell, the stream is bytes by default
copy-feats ark:$1 ark:-
# $1 is the mfcc-features-file or stream('-'), '-' means output data stream in the shell, the stream is text
copy-feats ark:$1 ark,t:-
# $2 is a file to save byte stream data.('ark,t' means saving text stream)
copy-feats ark:$1 ark:$2
# copy mfcc-feats-file to $ark(bytes or text) and $scp
copy-feats --compress=true ark:$1 ark,[t,]scp:$ark,$scp



compute-cmvn-stats scp:$f2.scp ark,scp:$test-cmvn.ark,$test-cmvn.scp
apply-cmvn scp:./test-cmvn.scp scp:./QT-742440ans.wav.scp ark,t:-


'''
./online2-wav-nnet3-latgen-faster --do-endpointing=false --online=false --feature-type=fbank 
	--fbank-config=../../egs/cvte/s5/conf/fbank.conf 
	--max-active=7000 --beam=15.0 --lattice-beam=6.0 --acoustic-scale=1.0 
	--word-symbol-table=../../egs/cvte/s5/exp/chain/tdnn/graph/words.txt 
	../../egs/cvte/s5/exp/chain/tdnn/final.mdl 
	../../egs/cvte/s5/exp/chain/tdnn/graph/HCLG.fst 
	'ark:echo utter1 utter1|' 'scp:echo utter1 ../../egs/cvte/s5/data/wav/00030/2017_03_07_16.57.22_1175.wav|' ark:/dev/null

online2-wav-nnet3-latgen-faster：是调用来进行识别的命令
后面跟着的那些是参数设置和输入的文件和模型
words.txt词表
final.mdl训练好的声学模型
HCLG.fst
H:HMM，输出符表示上下文相关的音素
C：表上下文关系，输出是音素，输入是上下文相关的N个音素组成的窗
L：lexicon发音词典，输出是词，输入是音素
G：语法
最后那个WAV是你要识别的语音。
'''