Data transmission via standard audio cable

Hi, I'm trying to transmit an image from a computer to another using Matlab and a standard aux cable. On the transmitter I get the binary matrix of the image in b&w with im2bw, convert it into an electrical signal (1 -1) and use the QPSK modulation, without using Matlab predefined functions (just by multiplying half of the bits for cos(2*pi*f*t) and the other for sin(2*pi*f*t) ). Then I use sound while I use audiorecorder on the receiver, so that I get data on the other pc and then I make the demodulation.

This said, here it is an example of my results:

Original file

Transmitted image

Received image without filters

Received image with a low pass filter (not sure it's well done)

Received image with a band pass filter (not sure it's well done)

The obvious question is how to receive the correct image?

More in detail,

1) can you suggest me a good filter (with matlab code)? I'm using it before demodulating the received signal, is it right?

2) I'm "cheating" on reconstructing the image, because I already know what the specific dimensions of this particular image are (width and height), but if my transmission would be far more noiseless, I would transmit the dimensions before transmitting data. Do you think I can solve this problem using a FEC method like Hamming or Reed Solomon?

3) I see Mickey's head is cut off on the received images, and I think the problem is due to the delay between the start of recording and the start of transmission, but if you help me with point 2) (increasing the precision) I should be able to say to the receiver what is the beginning of data by means of a pattern

4) Any other suggestion is really appreciated!

Thanks a lot to everyone who tries to help me

[srikanth GS]

Interesting, please check the filter bandwidths of both the sides

[Walter Roberson]

I would suggest that you send swaths, each of which has a header that identifies the size of the image and the position of the swath within the image. Yes, definitely use error detection and correction.

[yanqi liu]

may be use base64 encode and decode to make image to string and bits，then transfer it

[Walter Roberson]

@yanqi liu

They are already using im2bw() so they only have bits anyhow.