Heard of it - I might give it a try. But usually MCompare does the job for me, also from the screenshots there seems to be a "blind test" modules as well in MCompare (haven't tried it though). Maybe I'll give that one a shot first, and report back...
Cheers,
Codex
MTurboEQ
-
- KVRist
- 243 posts since 17 Sep, 2006
-
- KVRist
- 243 posts since 17 Sep, 2006
Ok did the tests, there you go...
MTurboEQ - Moto Town Plus -Airband Gain +10dB with 1x Oversampling vs.
MTurboEQ - Moto Town Plus -Airband Gain +10dB with 2x Oversampling vs.
MTurboEQ - Moto Town Plus -Airband Gain +10dB with 3x Oversampling
1) Blind Test in MCompare: In 4 out of 5 times I chose the 2x or 3x over the 1x
2) Null Test:
- Base Line: When both channels where set to 1x oversampling the signals nulled out as expected (MLoudness showed -145dBFS)
- 1x vs. 2x: MLoudness showed a residual of -59dBFS and the residual was very clearly hearable with my normal listening volume
- 1x vs. 3x: MLoudness showed a residual of -58dBFS, still very clearly hearable but no perceivable difference to 1x vs. 2x
Discussion:
First, there is definitely a difference between 1x and 2x in the null test. Although -59dBFS does not seem a lot the residual was clearly hearable at the same listening volume as I listened to the un-nulled signal, so the residual is in the perceivable loudness range. Moreover in the blind-test I chose the upsampled signal more often than the normal one. (However statistically I would need a lot more trials to get significance)
Cheers,
Codex
MTurboEQ - Moto Town Plus -Airband Gain +10dB with 1x Oversampling vs.
MTurboEQ - Moto Town Plus -Airband Gain +10dB with 2x Oversampling vs.
MTurboEQ - Moto Town Plus -Airband Gain +10dB with 3x Oversampling
1) Blind Test in MCompare: In 4 out of 5 times I chose the 2x or 3x over the 1x
2) Null Test:
- Base Line: When both channels where set to 1x oversampling the signals nulled out as expected (MLoudness showed -145dBFS)
- 1x vs. 2x: MLoudness showed a residual of -59dBFS and the residual was very clearly hearable with my normal listening volume
- 1x vs. 3x: MLoudness showed a residual of -58dBFS, still very clearly hearable but no perceivable difference to 1x vs. 2x
Discussion:
First, there is definitely a difference between 1x and 2x in the null test. Although -59dBFS does not seem a lot the residual was clearly hearable at the same listening volume as I listened to the un-nulled signal, so the residual is in the perceivable loudness range. Moreover in the blind-test I chose the upsampled signal more often than the normal one. (However statistically I would need a lot more trials to get significance)
Cheers,
Codex
-
- KVRist
- 243 posts since 17 Sep, 2006
Oh and for comparison:
When you null
MTurboEQ - Moto Town Plus -Airband Gain +0dB with 1x Oversampling
vs.
MTurboEQ - Moto Town Plus -Airband Gain +10dB with 1x Oversampling
the residual shows a loudness of -32dBFS. That means that this 10dB gain by itself only leads to a difference of -32dBFS in the residuum, which puts the -59dbFS of the former tests into the right scale.
Dammit guys, the Moto Town Plus -Airband in 2x mode is nothing short of impressive. I tried it on different material, voices mainly, and it adds such silky smooth high-frequencies to it! I'd say "Go home Mäag Audio EQ4", but I don't have it, so this would be made up of thin air...
Please try it for yourselves and then tell me if I went completely crazy!
When you null
MTurboEQ - Moto Town Plus -Airband Gain +0dB with 1x Oversampling
vs.
MTurboEQ - Moto Town Plus -Airband Gain +10dB with 1x Oversampling
the residual shows a loudness of -32dBFS. That means that this 10dB gain by itself only leads to a difference of -32dBFS in the residuum, which puts the -59dbFS of the former tests into the right scale.
Dammit guys, the Moto Town Plus -Airband in 2x mode is nothing short of impressive. I tried it on different material, voices mainly, and it adds such silky smooth high-frequencies to it! I'd say "Go home Mäag Audio EQ4", but I don't have it, so this would be made up of thin air...
Please try it for yourselves and then tell me if I went completely crazy!
-
MeldaProduction MeldaProduction https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=176122 - KVRAF
- 14019 posts since 15 Mar, 2008 from Czech republic
For the record, choosing options 2x or 3x instead of 1x 4 out of 5 times is statistically a horrible result proving nothing 
. (my wife is a statistician ) Try it 20x, if you will choose 2x or 3x at least 17 times, then it is a reasonable result. Or just test 1x & 2x and if you choose 15 times 2x, then it is a usable result. Otherwise the statistics is just not usable at all.
. (my wife is a statistician ) Try it 20x, if you will choose 2x or 3x at least 17 times, then it is a reasonable result. Or just test 1x & 2x and if you choose 15 times 2x, then it is a usable result. Otherwise the statistics is just not usable at all.-
- KVRist
- 243 posts since 17 Sep, 2006
I know, I know... I had a pretty good statistics education in university myself. The probability is about 0.46 to choose it 4 out of 5 times (or more).
17 out of 20 (or more) would be an alpha-error of 6%, so pretty close to statistical significance, yes...
17 out of 20 (or more) would be an alpha-error of 6%, so pretty close to statistical significance, yes...
-
MeldaProduction MeldaProduction https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=176122
- KVRAF
- 14019 posts since 15 Mar, 2008 from Czech republic
Interesting! Well the filter shapes can be slightly different with different sampling rates, but I'm surprised anyways!
Well the filter shapes can be slightly different with different sampling rates, but I'm surprised anyways!
