[opus] Phase perservation in Opus

[Jean-Marc Valin]

Hi,

You have just failed to take into account the fact that Opus introduces
a fixed delay. Put in a more complex signal and you will see that it's
not a phase change but a simple delay due to the look-ahead of the
encoder. All (non-trivial) codecs do that.

	Jean-Marc

On 12/04/2013 04:53 AM, Bent Bisballe Nyeng wrote:
> Hi list
> 
> I'm writing some unit tests for a piece of software using the opus 
> library for audio encoding before transfering though RTP.
> In order to test my RTP implementation I thought I would transfer an 
> encoded 440Hz sinus, and then do a diff of the received/decoded signal.
> However, my strategy seems to fail completely since the recreated sinus 
> looks nothing like the one in the decoded signal.
> 
> So my question is; is the phase not likely to be preserved during opus 
> encoding/decoding or did I simply make an error in my test code?
> 
> I am aware that the diff should not be binary equivalent to the input 
> signal, but my thought is to do an error calculation and accept the test 
> if under a certain threshold.
> 
> Kind regards
> Bent Bisballe Nyeng
> _______________________________________________
> opus mailing list
> opus at xiph.org
> http://lists.xiph.org/mailman/listinfo/opus
>