command and control

[rastinrastini]

Hi
i now train an data with text and audio with near 900 audio file. but its acc is low.
now i want to learn how can do command and control program but dont understand tutorial.
what is steps? below step is my steps for data training. please correct it for command and control.
1- prepare text data with one sentence in each row.
2- setup app with command: "sphinxtrain -t an4 setup"
3- create vocab data and convert ext to .dict.
4- create lm file with this command: ngram-count -kndiscount -interpolate -text train-text.txt -lm your.lm
5- split chars of dict file in front of word in each line like: "word w o r d"
6- create transcription file with one sentence in s tag in each line and number of audio related.
7- create fileids file with audio name in each line.
8- create phone list file.
9- edit cfg file.
10- train data with command: "sphinxtrain run"

what must i do to create command control app?
Thankful.

[Nickolay V. Shmyrev]

now i want to learn how can do command and control program but dont understand tutorial.

There are many courses which teach you how to create software, for example this one:

https://www.coursera.org/learn/learn-to-program

Basically you open text editor and start writing code, then test it, package and distribute.

below step is my steps for data training. please correct it for command and control.

Steps for creating application and acoustic model training are not really related.

[rastinrastini]

sorry. im developer. i can create application. what can i do for creatign command and control app.
it means what works must do?
create file with words in each line?
step by step how can do it and prepare for app to use it?

[Arseniy Gorin]

If you work with English, no need to train new acousic model. Download it from sourceforge. Then you need your own dictionary and language model (4-5 in your plan above except in 5 you should take phonemes, not characters. You better take pronunciations from sphinx dictionary). Then you run decoding and it's done.
However, if your language has no acoustic model supplied with sphinx, training is another story and you should really carefully study the tutorial in this case

[rastinrastini]

i studied but dont understand how can for grammer. i dont use english.

then how can config grammer?
below is my config but dont work with it and return another words.

configuration
.setAcousticModelPath("file:" + ws);
configuration
.setDictionaryPath("file:" + ws1);
configuration
.setLanguageModelPath("file:" + ws2);
configuration.setGrammarPath("file:F:\projects\speech\test_speech\test_speech\src\main\java\com\mycompany\test_speech\hello.gram");

[rastinrastini]

if i want 4 command then what must do from start?
words: "word1 word2 word3 word4"

[rastinrastini]

if i want cleaner:
i know must create gram file with jsg format. but dont know what must do next.
do you know?