Who can tell where I can download the test audio sequences which are used in ITU Recommendation BS. 1387, I cannot find them by internet, except for acodsna.wav and arefsna.wav. Thank you very much.

I think that they are not publicly available.

Regretful, I just want to validate my implementation of PEAQ.
BTW, my host can get the test sequences from ITU, but I have not found the link of these sequences in ITU. Who know the site? Thanks a lot.

Please don't forget to post here in case that you get the same results as those stated in ITU-R BS.1387. In my opinion something is missing in ITU-R BS.1387-1 (you can check this with EAQUAL and PQEvalAudio) or ITU implementation of PEAQ is not implemented as it is described in ITU-R BS.1387-1.

One correction to my previous post: Samples to check conformance of an implementation of ITU-R BS.1387-1 are available on CD that you can purchase from ITU. Complete set of samples that was used to develop ITU-R BS.1387-1 is not available to the public .

Thank you very much for the information, I just want to implement the basic version of PEAQ, since I am a beginner of audio quality. Our goal is to construct the objective audiovisual quality metrics. If I can get the same results, I will post it.

Hi!
How is going on your implementation?. I´m trying the same thing as you, a Basic Version of this Rec.
Well, first of all, excuse my english!

Had you found the signals?. Here you have an email where you can buy the CD that includes them.

[A CD ROM associated with this Recommendation (which contains the WAV-files referenced in section 7.3) can be purchased separately from the ITU Sales department (Email: sales@itu.int)]

Of course, the Recomendation is not clear in many aspects. P. Kabal´s document was very helpfull with his PQevalAudio aplication. But what about EAQUAL?. Is this the german C++ implementation?. If not, which one is?.

Well, Ihave only acodsna, arefsna, bcodtri, breftri, ccodsax, crefsax, ecodsmg, erefsmg but maybe they can be helpfull for you. Tell me.

In which language are you programming?. I´ve tryed in Matlab and then in C#, the problem is it has not many good libraries with good math´s functions.

Do you know somethig more about the calibration of the input signals (alignment, attenuation...)?. It´s not clear for me.

And do you have the ITU implementation?. I have not it. How can I get it?.

You posted the messages two weeks ago, If it´s not very late, we could discuss some aspects.

Thank u!

The ITU implementation you will not find. And the closest implementation to it is Opticom Opera which costs a lot of money and comes without source code.
You can however find at least 3 publicly available implementation of Basic model of ITU-R BS.1387 which produce very similar output (but different from Opera and from ITU reports, but even Opera has some small differences from ITU reports as I remember).

I am sorry to reply later, I have been very busy recently.

I have completed the implementation of the Basic Version, and I got the similar results comparing to McGill Univ. and EAQUAL, but some results do not meet the ITU tolerance requirement.

Thanks a lot for your information. In fact, I don't know which language of EQUAL, perhaps C. My implementation is completed by Matlab. P. Kabal´s document can provide much help, and my routine are a little slow than theirs, maybe I need to optimize them.

I have also made some experiments for other sample rates and frame setting, unfortunately, the results are not stable.

If you get some interesting results, we can discuss about them, thank you.

BTW, I don't think the calibration for the input signals is needed, because they had been calibrated, I mean the standard test sequences provided by ITU. (I got them from NokiA).