You if you are sending raw image/video or audio (in pcm) then you can calculate my applying simple maths below:

Constant bitrate formula: Uncompressed data rate = color depths * vertical resolution * horizontal resolution * refresh frequency

Examples

24bit @ 1080i @ 60fps :24*1920*1080*60/2=1.49 Gbps.

24bit @ 1080p @ 60fps :24*1920*1080*60=2.98 Gbit/s.

You can check the links:

http://web.forret.com/tools/video_fps.asp

http://en.wikipedia.org/wiki/Uncompressed_video

But if you are taking about video band width or audio bandwidth after compression using a codec like H.264 (for video ) or G.721 for (audio), the the bit rate solely depends on the internal codec design where I P B compressed frames come in to play for video and Codebook come into play for Audio G.721 codec( giving 4-3 kbps rate)

It also depends on the vbr or cbr mode you are using.In the first cast there is variable bit rate and for cbr constant bit rate can be observed.

Currently we can achieve about 1mbs for 720p for streaming using H.264 codecs and about 4-8 kbps using G.729 codec.(for exact values u can check internet or you can stream a video or audio using VLC and check the bandwidth consumption using take manager networking tab,

The usual definition of bandwidth is a difference between two frequencies chosen to encompass some frequency region of interest.  For example, a band pass filter has a bandwidth typically described as follows.  Define the center frequency, Fc, as the frequency most readily transmitted by the band pass filter.  Now locate two frequencies, as close as possible to Fc, where the response of the filter is 3 decibels (dB) less than the response at Fc (the so-called "half-power points").  The bandwidth is then the difference between these two frequencies.

Another definition is the difference between the highest and lowest frequency used anywhere in a given signal.

In digital communications, bandwidth is also related to the speed of bit transmission.  Whether these bits represent physical computer memory or Shannon/Weaver-style information measures is a matter of choice.  For example, a standard audio CD uses 44,1100 samples per second to represent audio.  Since there are 2 audio channels (for stereo) and 16 bits per sample, the total bandwidth is 2 x 16 x 44100 = 1,411,200 bits per second.  Real communication channels typically have additional requirements for headers and "side information" such as lyrics.  This is also called the data transmission rate.

But if the data bits are all zero, the amount of information transmitted through a channel of any capacity can be zero as well.

Following the explanation of Dr. Moore, you should obtain a sample of the signal and obtaining the power spectrum (square of the module of the fast fourier transform). After that, you can fix a thresold of the amplitude that you take as signal and calculate the two frequencies of the bandwith. Good luck!