Codec Landscape
Quality vs Bitrate
The figure below illustrates the quality of various codecs as a function of the bitrate. It attempts to summarize results from a collection of listening tests and (when no data exists) show anecdotal evidence. It is overall fairly representative, but attempting to extract any exact value at a particular bitrate is certainly not recommended.
Bitrate/Latency Comparison
Listening Tests
Several tests were conducted on Opus, but only the ones conducted on the final bit-stream are listed below. Although these should give a good idea of the quality of Opus at the time of its standardization (and 1.0 release), we are hoping that newer and more advanced encoders will reach even better quality.
HydrogenAudio (64 kb/s)
These are the results of ABC/HR tests comparing Opus with Vorbis and HE-AAC on 48 kHz stereo music at 64 kb/s. For the full details, see the official results page and Greg Maxwell’s analysis/listening page.
HydrogenAudio (96 kb/s)
These are the results of a second set of ABC/HR tests on 48 kHz stereo music comparing Opus with Vorbis and LC-AAC at 96 kb/s and MP3 at 136 kb/s. For the full details, see the official results page.
Google listening tests
Jan Skoglund from Google organized two sets of listening tests. The first set of listening tests includes a narrowband speech test, a wideband-fullband speech test, and a stereo music test.
The second set of listening tests measures the narrowband and wideband/fullband speech quality on Mandarin and the transcoding quality in narrowband and wideband.
Nokia listening test
Anssi Ramo and Henri Toukomaa from Nokia measured the Opus speech quality at various rates and published their results in this conference paper:
- Anssi Ramo and Henri Toukomaa, Voice Quality Characterization of IETF Opus Codec, Proc. Interspeech 2011, Florence, Italy, August 2011.
In this paper, the Opus “hybrid” and “MDCT” curves measure the constant bitrate (CBR) quality, which is not as good as the quality Opus achieves with variable bitrate (VBR).