To know exactly how many bytes a voice conversation consumes, you need to know which codec your VoIP service is using. A codec is a compression engine that transforms your (analog) voice into digital data, removing the silent moments (which make up to half of all conversations), and do other things to render the data load as light as possible. Read more on
codecs there.
Here are approximate values for data consumption of the most common codecs used for VoIP:
G.711 - 87Kbps
G.729 - 32 Kbps
G.723.1 - 22 Kbps
G.723.1 - 21 Kbps
G.726 – 55 Kbps
G.726 – 47 Kbps
G.728 - 32 Kbps
These values will give you matter for calculation. For example, for one minute of talk with the G.729 codec, we will do the following calculation:
G.729 takes 32 kilobits per second,
which is 1920 kilobits (60 x 32) in one minute,
which in turn is 240 kilobytes (KB) per minute (1 byte is 8 bits)
Now that’s only for the data going out. Inbound data (which also counts) takes the same load, so we double the figure to 480 KB.
Finally, we can round the value to 0.5 MB per minute of conversation.
The G.729 codec is one of the best performing voice codecs and most good VoIP services use it.
You should note that there are many parameters, that are rather technical in nature, affecting the values above. Among them are the size (payload) of the voice packets, the intervals at which they are sent and the number of packets sent in one second (frequency). For most of us, what we want is an approximation for an estimate. So, we can easily do away with the accuracy. Also, we might not know which codec is being used. Personally, I take the average value of 50 kbps for any codec. This gives (after calculations and approximations) 0.75 MB per minute of conversation.
So, if you plan an hour of conversation, it will be roughly 45 MB.