[open] Audio threshold

snufkin

Hi!
I am trying for long time now to detect an audio threshold from the microphone through the inline code.
The problem is that I am trying to do that in a loop, I am opening the stream each loop which is time consuming and sometimes doesn't work…

Is there any way to open the microphone stream once in the start of the experiment (exp.) and using the data when needed???

The michrophone code:

TAP_THRESHOLD = self.experiment.get("mic")
FORMAT = pyaudio.paInt16 
SHORT_NORMALIZE = (1.0/32768.0)
CHANNELS = 2
RATE = 44100  
INPUT_BLOCK_TIME = 0.01
INPUT_FRAMES_PER_BLOCK = int(RATE*INPUT_BLOCK_TIME)

def get_rms( block ):

    """Get root mean square as a measure of loudness"""

    import struct
    import math 

    global SHORT_NORMALIZE

    count = len(block)/2
    format = "%dh" % (count)
    shorts = struct.unpack( format, block )
    sum_squares = 0.0
    for sample in shorts:
        n = sample * SHORT_NORMALIZE
        sum_squares += n*n
    return math.sqrt( sum_squares / count )

# Open the mic
try:
    stream = pyaudio.PyAudio().open(format = FORMAT,
    channels = CHANNELS,
    rate = RATE,
    input = True,
    input_device_index = 0,
    frames_per_buffer = INPUT_FRAMES_PER_BLOCK)
except IOError, e:
    print e
    print "mic stream error1"

sebastiaan

Hi,

In principle, creating and closing a stream for every trial should be fine. From you code snippet it's not clear how you are trying to do this exactly. Are you sure that you are closing the stream? Working examples can be found in this discussion:

• http://forum.cogsci.nl/index.php?p=/discussion/6/solved-voicekey-input-triggering-responses-through-sound-input

If the issue persists, could you remove the try ... except ... from your code and post the stacktrace from the debug window? The error message should give some insight into what goes wrong.

Alternatively, I suppose you could create a stream object at the start of the experiment (in the same way as you're doing now), and call stream.start_stream() at the start of the trial, and stream.stop_stream() at the end. For documentation, see:

• http://people.csail.mit.edu/hubert/pyaudio/docs/#pyaudio.Stream.start_stream

But I would think that simply creating a new stream on every trial should work just as well.

Cheers!
Sebastiaan

snufkin

hi sebastiaan
Thanks for the quik answer

It seems that sometimes the pyaudio gets an garbage as an input, very high values from the mic.

Here is the complete code

import pyaudio 
import serial
import time
from time import sleep 
from openexp.keyboard import keyboard
time1=self.time()

flagn=0 # this flag is the 'r' option flag that will stop the rotation 
my_keyboard = keyboard(exp, keylist=['escape'], timeout=1)
my_keyboard2 = keyboard(exp, keylist=['1','2','3','4','5','6','7','8','9','0',None], timeout=1600) #1/3this should be changed according to how much time we need after each vibration in miliseconds

# fixation point
from openexp.canvas import canvas
my_canvas = canvas(exp)
my_canvas.set_penwidth(8) #penwidth
my_canvas.line(410, 390, 610, 390)  #here you can change the fixation
my_canvas.line(510, 290, 510, 490)
my_canvas.show()
#self.sleep(1000)
#end fixation point



#ser = serial.Serial("com30", 9600)
time.sleep(2)
exp.ser.bytesize = serial.EIGHTBITS





in1 = self.experiment.get("f1")
in2 = self.experiment.get("f2")
in3 = self.experiment.get("f3")
in4 = self.experiment.get("f4")
in5 = self.experiment.get("f5")
in6 = self.experiment.get("f6")
in7 = self.experiment.get("f7")
in8 = self.experiment.get("f8")
in9 = self.experiment.get("f9")
in10 = self.experiment.get("f10")


vol = self.experiment.get("volume")

modevt = self.experiment.get("mode")

timeop = int(self.experiment.get("timemili")/10)


regb = 0
regd = 0
regb = in6 + in7*2 + in8*4 + in9*8 + in10*16 + 32

regd = in1*1 + in2*4 + in3*8 + in4*16 + in5*32



if modevt == 'a':
    exp.ser.write('a')
    time.sleep(0.025)
    #exp.ser.close()


#the beep code
from openexp.sampler import sampler
#src = exp.get_file('C:\Program Files (x86)\OpenSesame\sounds\pi.wav')
src = exp.get_file('pi.wav')
my_sampler = sampler(exp, src)
my_sampler.play()
#bip = self.experiment.get("beeptime")
bip = 200
self.sleep(bip)
#end of beep code


#microphone def

if modevt == 't':
    exp.ser.write('t')
    exp.ser.write(chr(regb))
    exp.ser.write(chr(regd))
    exp.ser.write(chr(timeop))
    exp.ser.write(chr(vol))

    time.sleep(0.01)
    #exp.ser.close()


if modevt == 'r':
    flagn=1
    exp.ser.write('r')
    exp.ser.write(chr(regb))
    exp.ser.write(chr(regd))
    exp.ser.write(chr(vol))

    start_response_interval = self.time()

    # Listen until a sound has been detected

#TAP_THRESHOLD = 0.1
TAP_THRESHOLD = self.experiment.get("mic")
FORMAT = pyaudio.paInt16 
SHORT_NORMALIZE = (1.0/32768.0)
CHANNELS = 2
RATE = 44100  
INPUT_BLOCK_TIME = 0.01
INPUT_FRAMES_PER_BLOCK = int(RATE*INPUT_BLOCK_TIME)

def get_rms( block ):

    """Get root mean square as a measure of loudness"""

    import struct
    import math 

    global SHORT_NORMALIZE

    count = len(block)/2
    format = "%dh" % (count)
    shorts = struct.unpack( format, block )
    sum_squares = 0.0
    for sample in shorts:
        n = sample * SHORT_NORMALIZE
        sum_squares += n*n
    return math.sqrt( sum_squares / count )

# Open the mic
try:
    stream = pyaudio.PyAudio().open(format = FORMAT,
    channels = CHANNELS,
    rate = RATE,
    input = True,
    input_device_index = 0,
    frames_per_buffer = INPUT_FRAMES_PER_BLOCK)
except IOError, e:
    print e
    print "mic stream error1"

normal_exit = 0
esc_exit = 0

start_response_interval = self.time()
# Listen until a sound has been detected
maxtime=self.time()
while True:
        try:        
            my_keyboard.get_key()
            #if response == None:           
            block = stream.read(INPUT_FRAMES_PER_BLOCK)
            loudness = get_rms( block )
            if loudness > TAP_THRESHOLD:
                print "mic exit"
                normal_exit = 1             
                self.experiment.set("response_time", self.time() - start_response_interval)
                if flagn == 1 :
                    exp.ser.write('n')
                # Close the mic
                stream.stop_stream()
                stream.close()
                break   
            if self.time()-maxtime > 3000:
                normal_exit = 1             
                self.experiment.set("response_time", self.time() - start_response_interval)
                if flagn == 1 :
                    exp.ser.write('n')
                print "max time"
                stream.stop_stream()
                stream.close()
                break
        except IOError, e:
            print e
            print "mic stream error"
            if flagn == 1 :
                    exp.ser.write('n')
            self.experiment.set("response_time", "check if mic is connected")
            # Close the mic
            stream.stop_stream()
            stream.close()
            break
        except Exception:
            print "esc exit"
            if flagn == 1 :
                    exp.ser.write('n')
            self.experiment.set("response_time", "problem, escape was pressed during test")
            # Close the mic
            stream.stop_stream()
            stream.close()
            esc_exit = 1
            raise
            break




if normal_exit == 1: #normal exit and not exception
    end_time2 = self.time()
    my_keyboard2.flush()
    responsek, timeout2 = my_keyboard2.get_key()
    self.experiment.set("responsek", responsek)
    #print timeout2
    #print end_time2
    end_time22= self.time()
    if end_time22- end_time2 < 1599:
        #print timeout2
        print "working"
        eittime = float(float(0) - float(end_time2 - end_time22)/1000) #3/3this should be changed according to how much time we need after each vibration in seconds
        print "eittime = %.4f" %(eittime)
        time.sleep(1.6-eittime)

sebastiaan

Hmm, ok, but what exactly is the problem here? I took your previous post to mean that you occasionally get an error when you open a new stream. But now it sounds like the stream is opening fine, but you're getting funny-sounding input from the mic.

Could you be a bit more specific about what the problem is exactly? Please also include full error messages (from the debug window) if there are any.

snufkin

Hi
Thanks for the help, I really need it…
the problem is that some times when i open the stream it gets a value of 8000 from the stream,
and some times it is ok and stops just when I talk.

This code runs in a loop many times.
the problem shows up just on an opening of the stream, and never appeared on the first time that I open the stream.

this led me to the conclusion that if I'll open the stream just once outside of the loop it will be ok…

sebastiaan

the problem is that some times when i open the stream it gets a value of 8000 from the stream,

I'm afraid that you're still not being specific enough!

What exactly has a value of 8000, and where did you get this value from? Is 8000 the loudness returned by get_rms() that you printed to the debug window or something like that? If so, is 8000 even an unusual value, and, if it is, what are usual values? Also, if error messages are not the problem, why do you have so many try ... except ... statements in your code? These generally just hide problems and make scripts more difficult to debug, so unless there is a good reason I would remove them.

You really need to provide a detailed explanation of the problem for anyone to shed some light on it.

As an aside, if you haven't already done so, I would also record some microphone input with a tool like Audacity. If this looks normal, then you know at least that the microphone input itself is fine.

Cheers,
Sebastiaan