Difference between revisions of "Opus"

From MultimediaWiki
Jump to navigation Jump to search
 
Line 7: Line 7:
 
Open standard for low-latency, scalable lossy audio coding based on the speech-oriented [[SILK]] codec and the low-latency [[CELT]] codec. Opus can transition between a linear prediction codec at low bitrates and a transform codec at high bitrates and supports a hybrid mode for a small bitrate range. Opus defaults to a latency that's 1/10th that of earlier music codecs (20ms vs. 200ms), making it possible for use in VoIP, and yet performs better in listening tests.
 
Open standard for low-latency, scalable lossy audio coding based on the speech-oriented [[SILK]] codec and the low-latency [[CELT]] codec. Opus can transition between a linear prediction codec at low bitrates and a transform codec at high bitrates and supports a hybrid mode for a small bitrate range. Opus defaults to a latency that's 1/10th that of earlier music codecs (20ms vs. 200ms), making it possible for use in VoIP, and yet performs better in listening tests.
  
Note that Opus does not exceed the performance of [[CELT]] for music at bitrates over 75 kbit (http://www.hydrogenaudio.org/forums/index.php?showtopic=97913&st=0) which means that the unconditional merge of [[CELT]] into OPUS doubled the amount of code to be present, even for applications which do not benefit from the addition.
+
Note that Opus does not seem to perform better than [[CELT]] for music at bitrates over 75 kbit (http://www.hydrogenaudio.org/forums/index.php?showtopic=97913&st=0). This means that the unconditional merge of [[CELT]] into Opus doubled the complexity (both code size and CPU usage) even for applications which do not benefit from the addition.
  
 
== Links ==
 
== Links ==

Latest revision as of 04:17, 25 March 2013

Open standard for low-latency, scalable lossy audio coding based on the speech-oriented SILK codec and the low-latency CELT codec. Opus can transition between a linear prediction codec at low bitrates and a transform codec at high bitrates and supports a hybrid mode for a small bitrate range. Opus defaults to a latency that's 1/10th that of earlier music codecs (20ms vs. 200ms), making it possible for use in VoIP, and yet performs better in listening tests.

Note that Opus does not seem to perform better than CELT for music at bitrates over 75 kbit (http://www.hydrogenaudio.org/forums/index.php?showtopic=97913&st=0). This means that the unconditional merge of CELT into Opus doubled the complexity (both code size and CPU usage) even for applications which do not benefit from the addition.

Links