audio - Python offline speech recognition -

i'm doing , application fallowing:

1:if noise detected microphone, starts record audio, until no noise detected. after it, audio recorded wav file.

2:i have detect words on it. there only, 5 10 words detect.

so far, code first part (detect noise , record audio). now, have list following words: help, please, yes, no, could, you, after, tomorrow. need offline way detect if sound contains these words. possible? how can that? i'm using linux , there no way change operational system windows or use virtual machine.

i'm thinking use sound's spectrogram, create train database , use classifier predict. example, this spectrogram of word. technique use?

thanks.

you can use pocketsphinx python, install pip install pocketsphinx. code looks this:

import sys, os pocketsphinx.pocketsphinx import * sphinxbase.sphinxbase import *   modeldir = "../../../model" datadir = "../../../test/data"  # create decoder model config = decoder.default_config() config.set_string('-hmm', os.path.join(modeldir, 'en-us/en-us')) config.set_string('-dict', os.path.join(modeldir, 'en-us/cmudict-en-us.dict')) config.set_string('-kws', 'command.list')   # open file read data stream = open(os.path.join(datadir, "goforward.raw"), "rb")  # alternatively can read microphone # import pyaudio #  # p = pyaudio.pyaudio() # stream = p.open(format=pyaudio.paint16, channels=1, rate=16000, input=true, frames_per_buffer=1024) # stream.start_stream()  # process audio chunk chunk. on keyword detected perform action , restart search decoder = decoder(config) decoder.start_utt() while true:     buf = stream.read(1024)     if buf:          decoder.process_raw(buf, false, false)     else:          break     if decoder.hyp() != none:         print ([(seg.word, seg.prob, seg.start_frame, seg.end_frame) seg in decoder.seg()])         print ("detected keyword, restarting search")         decoder.end_utt()         decoder.start_utt() 

the list of keywords should this:

  forward /1e-1/   down /1e-1/   other phrase /1e-20/ 

the numbers thresholds detection