|
Post by Steve Forte Rio on Jul 16, 2014 17:29:57 GMT
Taras, we have discussed this early in this topic. The difference between two gain corrections is extremely low and way below of just noticeable difference that a human is capable to hear. Please see this part of discussion audiofolks.proboards.com/post/100/threadOk. Then we can use ABC/HR Java bult-in gain correction. There is no problems with it. kamedo2, ok, the translation will be done tomorrow. What about an e-mail account for receiving the results? Could you provide me an account on your domain (where the test page will be located)? It makes sense. I accept. About IRC: I will be free tomorrow after 18-00 GMT. So, IgorC, tell us a time when we all can join #hydrogenaudio room and discuss the last points, as you said.
|
|
|
Post by kamedo2 on Jul 17, 2014 12:35:54 GMT
IgorC, I support it and made the rule a bit strict, and added the updated rule to the page. - If you rank the low anchor at 5.0, your result of the sample will be invalid.
- If you rank the mid-low anchor at 5.0, your result of the sample will be invalid.
- If you rank the low anchor higher than the mid-low anchor, your result of the sample will be invalid.
- If you rank the reference worse than 4.5, your result of the sample will be invalid.
- If you rank the reference worse than 4.5 on four or more samples, all of your results will be invalid.
listening-test.coresv.net/Taras, the e-mail account is listening.tests.ha@gmail.com. The test page will be located in listening-test.coresv.net/. I'm going to be in the IRC in around 19:00-23:00 GMT. (4:00-8:00 AM in Japan). So I'm going to go to bed soon.
|
|
|
Post by Steve Forte Rio on Jul 18, 2014 19:38:39 GMT
Russian version of the test page is ready. Kemedo2, I have uploaded screenshots and java scripts to my server, so please inform me if there will be some changes. IgorC, I support it and made the rule a bit strict, and added the updated rule to the page. - If you rank the low anchor at 5.0, your result of the sample will be invalid.
- If you rank the mid-low anchor at 5.0, your result of the sample will be invalid.
- If you rank the low anchor higher than the mid-low anchor, your result of the sample will be invalid.
- If you rank the reference worse than 4.5, your result of the sample will be invalid.
- If you rank the reference worse than 4.5 on four or more samples, all of your results will be invalid.
Igor is right, we must weed out the "noise data". According to probability theory I propose to consider user's results as invalid if 50 or more percent of his results don't meet above terms. For example, if 4 of 8 samples contain reference signal ranked worse than 4.5, the whole results will be invalid. Or we can make the rule more strict and set the threshold lower than 50%. Could you send me a password for that e-mail? I'm going to be in the IRC in around 19:00-23:00 GMT. (4:00-8:00 AM in Japan). So I'm going to go to bed soon. I will be free on Sunday for all the day (from 07:00 to 20:00 GMT).
|
|
|
Post by kamedo2 on Jul 19, 2014 15:00:48 GMT
|
|
|
Post by kamedo2 on Jul 20, 2014 18:51:38 GMT
|
|
|
Post by kamedo2 on Jul 21, 2014 2:26:54 GMT
Taras,
we have revised three points. Please reflect the changes in the Russian page. 1. first text: The purpose is to test AAC, Vorbis and Opus at 96 kbps against a classic MP3 128 kbps, and find out which codec produces the best quality. 2. Use the latest discarding rule. 3. abchr07.png, and abchr11.png. Now the third group is 5.0-5.0, not 5.0-4.9. How can I send you a password for the mail?
|
|
|
Post by Steve Forte Rio on Jul 21, 2014 12:48:00 GMT
Kamedo2, done. Thank you.
|
|
|
Post by Allenovalo on Aug 13, 2020 12:52:12 GMT
Hurry up to look into loveawake.ru you will find a lot of interesting things
|
|