How to stream audio to speech api while recording ?

Hello guys, I'm trying to reduce the latency time in my project, so I have been
trying to send chunks of an audio while I record it, what I had tried until now was mix a code that sends a previously recorded audio by chunks:

def RecognizeSpeech(AUDIO_FILENAME,CHUNK_SIZE):
   client = Wit('MYTOKENHERE') 
   def wavIterator():
       wav = open(AUDIO_FILENAME, 'rb')
       chunk = wav.read(CHUNK_SIZE)
       while chunk:
           yield chunk
           chunk = wav.read(CHUNK_SIZE)
   resp = client.speech(wavIterator(), None,
           {'Content-Type': 'audio/wav', 'Transfer-encoding':'chunked'})

with this tutorial's code: https://indianpythonista.wordpress.com/2017/04/10/speech-recognition-using-wit-ai/
and made this frankstein:

def recReturnWavIterator(RECORD_SECONDS, CHUNK_SIZE, client):

    #--------- SETTING PARAMS FOR OUR AUDIO FILE ------------#
    FORMAT = pyaudio.paInt16    # format of wave
    CHANNELS = 2                # no. of audio channels
    RATE = 44100                # frame rate
    CHUNK = CHUNK_SIZE          # frames per audio sample
    #--------------------------------------------------------#
 
    # creating PyAudio object
    audio = pyaudio.PyAudio()
 
    # open a new stream for microphone
    # It creates a PortAudio Stream Wrapper class object
    stream = audio.open(format=FORMAT,channels=CHANNELS,
                        rate=RATE, input=True,
                        frames_per_buffer=CHUNK)
    print("Listening") 
    for i in range(int(RATE / CHUNK * RECORD_SECONDS)-1):
        # read audio stream from microphone
        data = stream.read(CHUNK)
        yield data
    print("Finished recording")
def RecognizeSpeech(CHUNK_SIZE):
    client = Wit('MYTOKENHERE')
    resp = client.speech(recReturnWavIterator(5,CHUNK_SIZE,client),None,{'Content-Type': 'audio/wav', 'Transfer-encoding': 'chunked'})
    print('Yay, got Wit.ai response: ' + str(resp))

wich does return always an empty text
Yay, got Wit.ai response: {'_text': None, 'entities': {}, 'msg_id': '7ed74ba3-698a-41a9-8158-8dd3857c3808'}

Is it possible to do something like that ? How ?
PS: Sorry, I am not very experienced with programming.

Start with the issue's RecognizeSpeech and recReturnWavIterator functions, then inspect the pywit client.speech entry point and its audio request expectations. Establish whether microphone chunks can be streamed in this form and define completion by obtaining a non-empty transcript or documenting the supported limitation.