Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[jmone]

I read some of the older threads where there was some discussion from Jim and Matt but I don't recall the feature making the cut.  Is there anyway of doing ABX testing in MC, maybe with a DSP Plug In??? eg Something that allows ABX post the DSP would be great... you could then prove that you can hear the impact of say the new "loudness" (or any DSP) feature.

If not, do I have to install something like foobar (or other prog??) to do simple ABX testing of say various bit rates VS lossless VS High Def (as I'm sure I've gone deaf since being on this forum!)

Thanks
Nathan

PS - I did not post on the open forums due to the mention of competitive products.

[mojave]

I see no mention of a competitive product! 

Here is how you can do something similar now. Use particles to create multiples of the same file. Change a tag to make each particle unique. Setup several zones with different DSP settings for what you want to test. Use ZoneSwitch to send each particle to a different zone based on the tags you previously setup. Select all your particles and send to Playing Now with shuffle turned on. Start playback of Playing Now.

If you only want to test two zones, create five identical particles for each zone so when they are shuffled you will have 10 chances of guessing which DSP is being used.

Each particle will play in its own zone with your DSP settings. Decide which you like better. Now go back to Playing Now and look at the shuffled list to see which file/DSP you liked best. You can also assign a playback range to the particles so you are comparing shorter portions.

If you move particles, make sure to re-edit their path by hand. 

[jmone]

Excellent Suggestion - and did I mention you are A funny man?  

I should also do the same to test if I can hear the difference between different bit rates and depths but for comaring formats (eg MP3 128kbps and FLAC I'd need to first make these tracks).

We are off for a few days now so well have a crack and report back.... Once I get some Hospital Grade AC Wall Plates that is...

[MrC]

This thread got me curious.

I have 3 zones, and created exclusive playback rules in each zone.  The first zone switch rule generates and saves a random value between 0 to 3.  This zone is targeted if the value is < 1.

The second zone uses the value generated above; if the value < 2, this zone is targeted.

The third zone is the fall-through case (i.e. values < 3, but >= 2).

I show all three zones in tabs, and in a 4th tab, show the Overview.

In a split pane, I have a single track which I double-click to start playback.  Often, MC will generate a "Playback Problem" dialog "Something went wrong with playback", but show the track is added to the targeted zone.

I don't think MC is correctly or robustly stopping playback in other zones and starting playback of a given track.  I also see that sometimes tracks appear to be playing in multiple zones, despite ZoneSwitch being configured to stop playback in all other zones.  Finally, periodically, MC will show a status of Waiting in the player bar but never progresses beyond that (no track playback, nothing happens until a track is double clicked again).

For the curious, the rules are:

Rule 1: [=save(math(rand(3)),R)math(below(load(R),1))]=1

Rule 2: [=math(below(load(R),2))]=1

Rule 3: [Media Type]=[Audio]

[mojave]

Does the global variable need to have the zone used at least once before a value is set?

[MrC]

It is evaluated when the zone switch's expression rule is run.  In any event, worst case, a rule that didn't work would default to the last zone rule in the list, and that should not cause these issues.

[Matt]

Here's a way I often do comparison tests:
http://yabb.jriver.com/interact/index.php?topic=74999.msg508326#msg508326

[jmone]

Reading the foobar thread one feature is that it records your preferences so you can not cheat.  Imagine hoe much fun such a feature would generate on interact.  You could have Goldern ears arguing with deaf ones like me

[jmone]

I thought I would start with something easy!  Boy am I surprised!  

I took a HiRes 88.2khz 2.0 24-Bit audio 94MB FLAC track (Police King of Pain) and got MC to convert two additional MP3 Versions, a Small Portable (4MB at 107Kbps), and an Extreme ( 9.5MB at 259Kbps) using the LAME encoder.  I then re-Analysed Audio on all three.

I then listened to all three on my SR60 Grado Labs Headphones from my FiiO DAC and cracked up the vol.

The good news was I can defiantly pick the Small Portable version.  The bad news I think it sounds better as the background hiss has been removed allowing more of the softer detail to be heard.  It is sort of like the old school tape based Dolby NR kicking in.  I can not reliably tell the difference between the Extreme and the original FLAC version.

I must be defective... or the equipment is... or the master is bad... or ... something as this can not be true!  Surely!  If it is I'm flogging all my gear off and getting an iPod!

Yours in depression
Nathan

EDIT - made another set of test files but will have to compare tommorow.  Now I know why MC does not include an ABX module!

[InflatableMouse]

Haha! 

It can be quite an eye-opener for sure.

I've organised several ABX tests on different audio sets with different people using Ogg, mp3, flac and Wav files. Ripped 5 or 6 songs from different genres with EAC (with that md5 or hash plugin), used command line flac, ogg and lame encoders to encode each song to 3 bitrates for Ogg and Mp3. Encoded them all back to wav and burned to 2 CDs. Clustered the songs together with the first one always being the Wav, the rest random and printed out sheets for the listeners to record their results. Funny observation was that flac and wav did not get a 100% score and was sometimes believed to be an ogg or mp3. This just goes to show how your brain plays tricks on you when you're set on hearing things, you simply hear them even though its not there.

Basically it came down to Ogg being transparent at 160kpbs, Mp3 at 192kbps (VBR). 128kbps for both Ogg and Mp3 got a 100% detection score, but what was interesting is that certain songs could be picked out even at the highest bitrates in Mp3 (I don't remember for sure but I don't think Ogg had that problem). This had to do with the soundstage. Some of the songs had a very defined, wide sound stage and mp3 messed that up. A particular detail that would appear to come from the far left in the original would suddenly sit closer to the middle with the mp3, or certain instruments would be less defined in terms of placement. Later when I listened to these tracks using headphones it did not reveal this detail (I had a pair of AKG's and a pair of Grado's as well).

The thing is, when people compare music they tend to focus on sound quality while the difference can be in the soundstage and when you're not listening to that specifically, I think its very easy to miss.

[Matt]

There are certain things you can learn to listen to that will sound different in a lossy copy -- a guitar string pluck, a high-hat symbol decay, etc.  

Focus on some background instrument as you listen to each copy.  Headphones at high volume are best.

Quiet or acoustic passages are more obvious because with something like Hard Rock it all blends together.

You need to listen to short passages (~10 seconds or less) back to back, because the human brain's memory for detail in sound isn't great.

Whether any of the differences you can hear matter is a different question.  Personally I got sick of trying to answer that question, so went with lossless.

[Supersnake]

The good news was I can defiantly pick the Small Portable version.

Please don't change the spelling

[jmone]

Please don't change the spelling

How prophetic 

[flac.rules]

I would love an ABX-function in MC, especially for testing DSP-settings.

[MrC]

The issue I mentioned in reply #3 is now resolved with build 118.

Here's the zone randomization playing back the same track - just keep hitting enter on a track and a random zone is selected.

[Frobozz]

An ABX testing component in MC would be fun, especially if it is able to make use of zones.  Making use of zones would allow for testing different DSP settings and also for things like testing two DACs and for other tests.

I tried to do an ABX test using Foobar to see if I could hear a difference between 16/44.1 and 24/96.  I was not able to get the test to be blind.  The problem was that my DAC would have a slight delay when switching sample rates.  That delay would be different when switching from 44.1 to 96 and different again when switching from 96 to 44.1 and different still when switching from 44.1 to 44.1 or 96 to 96.  Just by focusing on the delay during the ABX test I could figure out whether X was A or B.  The test became useless once I discovered how to cheat (it's impossible not to cheat once you know).

If MC had an ABX component that could take advantage of zones and if I had two of the same DAC.  I could set one zone and DAC to play 16/44.1 and the other zone and DAC to play 24/96.  I could route the output from each DAC to a mixer and then to a headphone amp.  Then maybe I could do a better test.