streaming pcm frame to server using pykandi.

[donaldos]

I want to perform frame-by-frame speech recognition using pykaldi.
I want to send voice data to the server with websocket every 250ms, and pykaldi will proceed with speech recognition after receiving pcm data from the server.
Please find the problem in the following code.
Thank you for your review.

using 0.2.2 version

1. convert pcm data to np array data

2. processing to recognize speech
   'def process_audio_chunk(audio_data, rate=16000):

   Convert audio to Kaldi format

   waveform = kaldi.matrix.SubVector(audio_data)

   Compute MFCC features

   mfcc_opts = kaldi.feat.mfcc.MfccOptions()
   mfcc_opts.frame_opts.samp_freq = 16000.0
   mfcc = kaldi.feat.mfcc.Mfcc(mfcc_opts)
   feats = kaldi.matrix.Matrix()
   mfcc.compute_features(waveform, rate, 1.0, feats)

   Perform decoding

   result = recognizer.decode(feats)'

3. problem.
   for mfcc.compute_feature(), please detailed example.
   in Documentation 0.1.1, argments is 4 but Error is happend in 0.2.2

[bmilde]

I recommend checking out kaldi-model-server (https://github.com/uhh-lt/kaldi-model-server). It's using pykaldi to do online decoding and it has good latency as well. You can see it in action here: https://ltdata1.informatik.uni-hamburg.de/meetingbot/meetingbot_de.mp4 (German) and https://ltdata1.informatik.uni-hamburg.de/meetingbot/meetingbot_en.mp4 (English).