[CELT-dev] measuring latency

[Jean-Marc Valin]

Quoting Tristan Matthews <tristan at sat.qc.ca>:
> I was just wondering what your setup was to measure latency for different
> encoders, i.e. lame vs. vorbis vs. celt mentioned here:
> http://jmspeex.livejournal.com/3839.html

These are *algorithmic* delays. They are not measured, but extracted from the
algorithm itself. The figures for Lame/Vorbis are very approximate though
because the encoders tend to use look-ahead in a rather complex way.

Cheers,

    Jean-Marc