Rhishikesh Agashe
2013-Oct-04 10:35 UTC
[opus] ODG (Objective Difference Grade) scores for Opus Encoder using PQEvalAudio Tool
Hi, I checked the ODG (Objective Difference Grade) scores for a few reference vectors using the PQEvalAudio Tool and found that some of them show ODG scores as high as -3.5 If we look at the range as described in the link below, it looks unacceptable. http://www-mmsp.ece.mcgill.ca/documents/Software/Packages/AFsp/PQevalAudio.html Am I missing something or are these scores valid? Thanks and Regards, Rhishi -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.xiph.org/pipermail/opus/attachments/20131004/7e809a4e/attachment.htm
Christian Hoene
2013-Oct-04 12:00 UTC
[opus] ODG (Objective Difference Grade) scores for Opus Encoder using PQEvalAudio Tool
Hi Rhishi, PQevalaudio is very unreliable and buggy. I have compared to PEAQ and - as a result - now I am not using it anymore. With best regards, Christian Hoene Von: opus-bounces at xiph.org [mailto:opus-bounces at xiph.org] Im Auftrag von Rhishikesh Agashe Gesendet: Freitag, 4. Oktober 2013 12:35 An: opus at xiph.org Cc: Rasmi Mishra Betreff: [opus] ODG (Objective Difference Grade) scores for Opus Encoder using PQEvalAudio Tool Hi, I checked the ODG (Objective Difference Grade) scores for a few reference vectors using the PQEvalAudio Tool and found that some of them show ODG scores as high as -3.5 If we look at the range as described in the link below, it looks unacceptable. http://www-mmsp.ece.mcgill.ca/documents/Software/Packages/AFsp/PQevalAudio.h tml Am I missing something or are these scores valid? Thanks and Regards, Rhishi -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.xiph.org/pipermail/opus/attachments/20131004/766bf06e/attachment-0001.htm
Rhishikesh Agashe
2013-Oct-04 12:01 UTC
[opus] ODG (Objective Difference Grade) scores for Opus Encoder using PQEvalAudio Tool
In that case, can you please suggest a reliable tool? Thanks, Rhishi From: Christian Hoene [mailto:christian.hoene at symonics.com] Sent: Friday, October 04, 2013 17:30 To: Rhishikesh Agashe; opus at xiph.org Cc: Rasmi Mishra Subject: AW: [opus] ODG (Objective Difference Grade) scores for Opus Encoder using PQEvalAudio Tool Hi Rhishi, PQevalaudio is very unreliable and buggy. I have compared to PEAQ and - as a result - now I am not using it anymore. With best regards, Christian Hoene Von: opus-bounces at xiph.org<mailto:opus-bounces at xiph.org> [mailto:opus-bounces at xiph.org] Im Auftrag von Rhishikesh Agashe Gesendet: Freitag, 4. Oktober 2013 12:35 An: opus at xiph.org<mailto:opus at xiph.org> Cc: Rasmi Mishra Betreff: [opus] ODG (Objective Difference Grade) scores for Opus Encoder using PQEvalAudio Tool Hi, I checked the ODG (Objective Difference Grade) scores for a few reference vectors using the PQEvalAudio Tool and found that some of them show ODG scores as high as -3.5 If we look at the range as described in the link below, it looks unacceptable. http://www-mmsp.ece.mcgill.ca/documents/Software/Packages/AFsp/PQevalAudio.html Am I missing something or are these scores valid? Thanks and Regards, Rhishi -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.xiph.org/pipermail/opus/attachments/20131004/63bfb054/attachment.htm
Jean-Marc Valin
2013-Oct-04 14:56 UTC
[opus] ODG (Objective Difference Grade) scores for Opus Encoder using PQEvalAudio Tool
On 10/04/2013 06:35 AM, Rhishikesh Agashe wrote:> I checked the ODG (Objective Difference Grade) scores for a few > reference vectors using the PQEvalAudio Tool and found that some of them > show ODG scores as high as -3.5What do you mean by "reference vectors"? The test vectors we provide or some of your own files you use for testing encoders. Regarding the testvectors, their quality is really bad (if you were to compare with the original) because they're designed to trigger as much as possible of the decoder code and not sound like the original. BTW, did you run all the testvectors with run_vectors.sh to see if the MIPS optimizations still produced a compliant decoder? Cheers, Jean-Marc