Show translation: exact meaning?


I am planning an Asterisk implementation for my company. I would like to size it right.

I will need to do lots of transcoding, as it will be connected to a legacy PABX. Therefore, the CODEC choice is of importance.

Using the SHOW TRANSLATION CLI command, I have the “time to translate” from a format to another.

But translate how much of the speech? Is it the time required to translate 1 frame, 1 second, 1 minute of speech in the original format?

If it states “20” ms to translate and if this is to translate 1 second of speech, can I consider I can have 50 (1000/20) channels in parallel to fill up my CPU (supposing the CPU only does that)?

Can someone define the exact meaning of the output of the SHOW TRANSLATION command?

Thanks in advance.


I think that time is the latency that will be introduced.

Thank you, Jen.

What do you mean when you say that you think it is the latency?

Are you about sure of it, of is it just how you think it might be operating?

I’m just trying to assess if you are positive about it.

Thanks again.


Jen2 is right - it is the number of milliseconds of latency that is introduced PER CALL for transcoding between the two codecs.

so if the number for gsm<->ulaw says 3, that means that there will be 3 milliseconds of latency introducted on each and every call that has to transcode between those two codecs.

Thank you Jen2 and Whoiswes for the responses…

Is there any way to assess the CPU load of a particular translation?

Best regards,


i don’t believe so, at least not within asterisk. you could probably come up with a load test of sorts…do a running load average, and ramp up 1, 2, 3…all the way up to 10 concurrent calls - you should be able to calculate the average CPU utilization increase for each call.

not very pretty, but it would probably give you a pretty good idea of what each translation costs in CPU time.

that would be my way of doing it, but i’m not a programmer either…