Synchronizing audio and video with OpenCV and PyAudio

Question

I have gotten both OpenCV and PyAudio working however I am not sure how I would sync them together. I am unable to get a framerate from OpenCV and measuring the call time for a frame changes from moment to moment. However with PyAudio it's basis is grabbing a certain sample rate. How would I sync them to be at the same rate. I assume there is some standard or some way codecs do it. (I've tried google all I got was information on lip syncing :/).

OpenCV Frame rate

from __future__ import division
import time
import math
import cv2, cv

vc = cv2.VideoCapture(0)
# get the frame
while True:

    before_read = time.time()
    rval, frame = vc.read()
    after_read  = time.time()
    if frame is not None:
        print len(frame)
        print math.ceil((1.0 / (after_read - before_read)))
        cv2.imshow("preview", frame)

        if cv2.waitKey(1) & 0xFF == ord('q'):
            break

    else:
        print "None..."
        cv2.waitKey(1)

# display the frame

while True:
    cv2.imshow("preview", frame)

    if cv2.waitKey(1) & 0xFF == ord('q'):
        break

Grabbing and saving audio

from sys import byteorder
from array import array
from struct import pack

import pyaudio
import wave

THRESHOLD = 500
CHUNK_SIZE = 1024
FORMAT = pyaudio.paInt16
RATE = 44100

def is_silent(snd_data):
    "Returns 'True' if below the 'silent' threshold"
    print "\n\n\n\n\n\n\n\n"
    print max(snd_data)
    print "\n\n\n\n\n\n\n\n"
    return max(snd_data) < THRESHOLD

def normalize(snd_data):
    "Average the volume out"
    MAXIMUM = 16384
    times = float(MAXIMUM)/max(abs(i) for i in snd_data)

    r = array('h')
    for i in snd_data:
        r.append(int(i*times))
    return r

def trim(snd_data):
    "Trim the blank spots at the start and end"
    def _trim(snd_data):
        snd_started = False
        r = array('h')

        for i in snd_data:
            if not snd_started and abs(i)>THRESHOLD:
                snd_started = True
                r.append(i)

            elif snd_started:
                r.append(i)
        return r

    # Trim to the left
    snd_data = _trim(snd_data)

    # Trim to the right
    snd_data.reverse()
    snd_data = _trim(snd_data)
    snd_data.reverse()
    return snd_data

def add_silence(snd_data, seconds):
    "Add silence to the start and end of 'snd_data' of length 'seconds' (float)"
    r = array('h', [0 for i in xrange(int(seconds*RATE))])
    r.extend(snd_data)
    r.extend([0 for i in xrange(int(seconds*RATE))])
    return r

def record():
    """
    Record a word or words from the microphone and 
    return the data as an array of signed shorts.

    Normalizes the audio, trims silence from the 
    start and end, and pads with 0.5 seconds of 
    blank sound to make sure VLC et al can play 
    it without getting chopped off.
    """
    p = pyaudio.PyAudio()
    stream = p.open(format=FORMAT, channels=1, rate=RATE,
        input=True, output=True,
        frames_per_buffer=CHUNK_SIZE)

    num_silent = 0
    snd_started = False

    r = array('h')

    while 1:
        # little endian, signed short
        snd_data = array('h', stream.read(1024))
        if byteorder == 'big':
            snd_data.byteswap()

        print "\n\n\n\n\n\n"
        print len(snd_data)
        print snd_data

        r.extend(snd_data)

        silent = is_silent(snd_data)

        if silent and snd_started:
            num_silent += 1
        elif not silent and not snd_started:
            snd_started = True

        if snd_started and num_silent > 1:
            break

    sample_width = p.get_sample_size(FORMAT)
    stream.stop_stream()
    stream.close()
    p.terminate()

    r = normalize(r)
    r = trim(r)
    r = add_silence(r, 0.5)
    return sample_width, r

def record_to_file(path):
    "Records from the microphone and outputs the resulting data to 'path'"
    sample_width, data = record()
    data = pack('<' + ('h'*len(data)), *data)

    wf = wave.open(path, 'wb')
    wf.setnchannels(1)
    wf.setsampwidth(sample_width)
    wf.setframerate(RATE)
    wf.writeframes(data)
    wf.close()

if __name__ == '__main__':
    print("please speak a word into the microphone")
    record_to_file('demo.wav')
    print("done - result written to demo.wav")

Answer 1

I think you'd be better off using either GSreamer or ffmpeg, or if you're on Windows, DirectShow. These libs can handle both audio and video, and should have some kind of a Multiplexer to allow you to mix video and audio properly.

But if you really want to do this using Opencv, you should be able to use VideoCapture to get the frame rate, have you tried using this?

fps = cv.GetCaptureProperty(vc, CV_CAP_PROP_FPS)

Another way would be to estimate fps as number of frames divided by duration:

nFrames  = cv.GetCaptureProperty(vc, CV_CAP_PROP_FRAME_COUNT)
           cv.SetCaptureProperty(vc, CV_CAP_PROP_POS_AVI_RATIO, 1)
duration = cv.GetCaptureProperty(vc, CV_CAP_PROP_POS_MSEC)
fps = 1000 * nFrames / duration;

I'm not sure I understand what you were trying to do here:

before_read = time.time()
rval, frame = vc.read()
after_read  = time.time()

It seems to me that doing after_read - before_read only measures how long it took for OpenCV to load the next frame, it doesn't measure the fps. OpenCV is not trying to do playback, it's only loading frames and it'll try to do so the fastest it can and I think there's no way to configure that. I think that putting a waitKey(1/fps) after displaying each frame will achieve what you're looking for.

Answer 2

You could have 2 counters 1 for audio and one for video. The video counter will become +(1/fps) when showing an image and audio +sec where sec the seconds of audio you are writing to the stream each time. Then on audio part of the code you can do something like While audiosec-videosec>=0.05: # Audio is ahead time.sleep(0.05)

And on video part While videosec-audiosec>=0.2:# video is ahead time.sleep(0.2)

You can play with the numbers

This is how i achieve some sort of synchronization on my own video player project using pyaudio recently ffmpeg instead of cv2.