I searched google, yahoo, and the forums here but could not find an answer.
I am going to have a system that does so many IVR’s that the file format is a huge issue.
What format requires the least amount of cpu cycles?
I was thinking 16bit sln, but I have nothing to back that up.
Centos/Fedora - 64bit
Core show translation will show you the cost of the each codec against your CPU.
Raw files (no .wav wrapper) in the codec used by the phone. You are allowed to have versions in multiple codecs. High compression codecs may cost less simply because there are less bytes to move, but only if they can be sent directly to the phone.
Thank you for your responses!
i believe that we go ulaw to the phones, so sln > ulaw shows 6000 in the translation matrix, which is the lowest.
The IVR will be talking to real landline phones (and potentially sip phones, and cell phones), over a trunk that uses ulaw.
Am I correct in assuming that using raw ulaw encoded files will be the least cpu intensive in this case? Is there anything special I need to do to make it skip the transcoding process (load file direct to phones) or will it do it automatically if it can?
[quote=“themrrobert”]
Am I correct in assuming that using raw ulaw encoded files will be the least cpu intensive in this case? Is there anything special I need to do to make it skip the transcoding process (load file direct to phones) or will it do it automatically if it can?[/quote]
If your incoming channel is a ulaw channel, then having file available in ulaw is the least computationally intensive, since there’s no transcode going from the source file format to the destination.