Decoder not working with Inputstream from TCP Client

[unrated]

I have a pythons tcp socket running on ubuntu and a java client running on windows. The java client sends an audio inputstream to the python application. The problem is im getting no hypothesis somehow from the decoder. At first i developed everything in python without a tcp socket. I used pyaudio to get the input stream from my microphone, i read the data from the stream and passed them to the decoder to process it. That worked very well. Now since im receiving a input stream from the client and pass the data into the decoder nothing is recognized at all.

Client
Audioformat for record:

static final AudioFormat DEFAULT_AUDIO_FORMAT = new AudioFormat(16000f, 16, 1, true, false);

public static void main(String[] args) {
        try {
            AudioRecorder ar = new AudioRecorder();
            ar.setAudioSystem(new ASWrapper());
            AudioInputStream is = ar.startRecording();
            Socket socket = new Socket("192.168.98.37", 9876);
            OutputStream os = socket.getOutputStream();

            byte buffer[] = new byte[512];

            while (true) {
                int read = is.read(buffer);
                ByteBuffer byteBuffer = ByteBuffer.wrap(buffer, 0, read);
                byteBuffer.order(ByteOrder.LITTLE_ENDIAN);
                os.write(byteBuffer.array(), 0, read);
                os.flush();
            }
        } catch (IOException e) {
            e.printStackTrace();
        }
    }

Server

import socket
import pyaudio
import wave
import time
import os
from pocketsphinx.pocketsphinx import *
from sphinxbase.sphinxbase import *

CHUNK = 4096
FORMAT = pyaudio.paInt16
CHANNELS = 1
RATE = 16000

WAVE_OUTPUT_FILENAME = "server_output.wav"
WIDTH = 2
frames = []

p = pyaudio.PyAudio()
stream = p.open(format=FORMAT,
                channels=CHANNELS,
                rate=RATE,
                output=True,
                input=True,
                frames_per_buffer=CHUNK)

HOST = ''                 # Symbolic name meaning all available interfaces
PORT = 9876               # Arbitrary non-privileged port

modeldir = "./"
datadir = "../../../test/data"

# Create a decoder with certain model
config = Decoder.default_config()
config.set_string('-hmm', os.path.join(modeldir, 'de-de/de-de'))
config.set_string('-dict', os.path.join(modeldir, 'de-de/train.dic'))
config.set_string('-kws', os.path.join(modeldir, 'de-de/kws.txt'))
#config.set_string('-lm', os.path.join(modeldir, 'de-de/train.lm'))
#config.set_string('-logfn', os.path.join(modeldir, '/dev/null'))

s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
s.bind((HOST, PORT))
s.listen(1)
conn, addr = s.accept()

decoder = Decoder(config)

print ('Connected by', addr)

data = conn.recv(CHUNK)
while data != '':
    # just to hear the sound quality
    stream.write(data)
    decoder.start_utt()
    decoder.process_raw(data, False, False)
    hypothesis = decoder.hyp()
    if hypothesis != None:
      print ('Best hypothesis: ', time.time(), hypothesis.hypstr, " model score: ", hypothesis.best_score, " confidence: ", hypothesis.prob)

    decoder.end_utt()
    data = conn.recv(CHUNK)

stream.stop_stream()
stream.close()
p.terminate()
conn.close()

[Nickolay V. Shmyrev]

Make sure you transfer data properly. Extra zeroes, byte order switch are very frequent mistakes. You can dump stream data and listen for it.