I wrote an AFSK encoder but am unhappy with it. I did the math to find

a sample rate where it would be an integer number of samples based on

the bitrate so the math is easy. I would prefer to be able to use an

arbitrary bitrate and sample rate instead.

I am wondering what the best practices method would be for these

transitional bits. For example let us assume I have the following

samples to send:

Space 20.3 samples

Mark 20.3 samples

Sample 21 is 0.3 space, 0.7 mark.

Sample 41 is 0.6 mark, 0.4 silence

What is the best way to encode those two samples without damaging the

output? I realize that with a high enough sample rate it would not

matter for most things, 44.1khz sample rate for example each sample is

only 0.02ms, but I would rather do it right than have garbage in the

output.

For sample 21 I am guessing (SPACE_FREQ*0.3)+(MARK_FREQ*0.7)

For sample 41 I am guessing (MARK_FREQ*0.6)

Is this correct? Is there anything that would be considered better/best

practice?

Thanks