2012/7/28 okay_awright <[email protected]>:
> hi,

Hi,

> can you please keep us posted on the performance gain of the whole MP3
> transcoding Liquidsoap-chain with and without Shine once you have
> rebuilt it?

On a pure C transcoding, shine performs 10-13x quicker than lame on
the raspberry pi. When run with liquidsoap on a simple transcoding
one-liner, it consumes about 50% of the CPU and 7% of memory.

However, liquidsoap's internals are all in floats, so if you add
smart_crossfade for instance, which computes RMS on the data, CPU
usage jumps to around 80% with a couple of catchups on track
transitions. The stream remains working though, thanks to buffering.

> A subjective listening test of this encoder, compared to Lame, would be
> really appreciated as well.
> The only qualitative test I could find was this one:
> http://listening-tests.hydrogenaudio.org/sebastian/mf-128-1/results.htm
> But raw results don't reflect whether it might sound good enough for a
> webradio.

I am personally not good at subjective testing. At least I remember
some contributors being much better than I am.

However, I personally find the resulting audio to be remarkably good.
I've tested a track with brass, drums and rich singing, which is
usually where I can hear artifacts and I found the encoded data to
sound really good.

Romain

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Savonet-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/savonet-users

Reply via email to