Welcome Guest ( Log In | Register )

potential mp3 difficulties, how important are these factors?
post Apr 8 2007, 22:49
Post #1

Group: Members
Posts: 2253
Joined: 31-August 05
Member No.: 24222

I've read a few contentions against lossy compression and another against many ABX test. I'm wondering if there is any established data to confirm or refute them, or if these factors have been taken into account in the development of LAME.

** The response of different speakers to the same power amplifier signals varies more or less everywhere across the frequency spectrum.
** The audio delivered to the ears from any given speaker system varies considerably from room to room due to room response and speaker-room interactions (not to mention the position of the listerner in the room).

Both of these are true and can be objectively measured. The contention is that psychoaccoustic models, where they are based on masking, are thus not completely valid except for the exact listening circumstance under which they were developed. This is because masking will differ somewhat under the different listening conditions, making transparency difficult or impossible to achieve in everyone's living room.

It seems well established that the listening room must be properly compensated to avoid adverse effects on what is heard. ‘Masking" of a type takes place at various frequency ranges in most "raw" rooms due to enhancements and cancellations from room reflections.

It is also well established that the majority of home music speakers, even rather expensive ones, are designed to deliver something other than exactly what is fed into them. This is why professional monitors exist.

The question is whether or not these factors are taken into account in encoding. If not, then the audiophile contention, that even the best mp3s are not transparent if the equipment is good enough, may have a basis in fact. Of course, in principal, ABX testing could be applied to investigating these claims. Have they? I suspect that most people who make such claims believe the differences are so apparent that further testing is irrelevant. It would probably take a skeptic -- with access to such equipment -- to make such tests.

The ABX testing "difficulty" contention is that the auditory system suffers from reduced discrimination ability after some time due to fatigue, and that this time can be as short as 20 minutes. A single ABX test probably most often takes less time, but more extensive tests probably run much longer. Is this a know factor? Is it considered relevant to encoder development?
Go to the top of the page
+Quote Post
Start new topic
post Apr 11 2007, 19:29
Post #2

Group: Members
Posts: 2253
Joined: 31-August 05
Member No.: 24222

All psychoacoustic-based lossy encoders are using the assumption that the output system frequency response should be flat.
My impression is that LAME development has been empirical, it has been fined tuned over and over based on ‘listening tests.' No?

Music is also mixed and mastered through listening tests, according to a well developed model. Professional monitors cost a lot because they are expensive to build, not because it is a snake-oil, whatever the market will bear, business. The near-field use of those monitors, in a properly treated room, is based on sound theoretical and empirical reasons. It is through development on this kind of system that the finished product is most likely to sound good on the widest range of real world playback systems.

I've seen many comments in this forum about what to use for one's personal testing. Comments along the line of ‘cheap earbuds that came with your walkman' have often been offered. For personal tests, to determine what encoding one wants to use, using whatever equipment you normally use to listen to music, makes good sense. For development of the codec itself, however, that does not seem reasonable.

Most people have no means to set-up a good near-field monitoring situation, nor a ideal living room listening setup. However, since there are significant differences in how at least some things sound on such systems, it seems to me that critical development would have to be done on a good monitoring setup. The only way anyone could reasonably claim otherwise is by collecting enough data, for enough people, using both "proper" and "other" setups, to support the nay contention.

I have no idea what in fact has been done with LAME development, that is why I ask. I am not saying LAME is deficient, I am merely asking how much these particular factors have been considered and proved. Is there data to say they are indeed well provided for or is there an underlying assumption that, since the majority of people will never hear recorded music played back under "proper" conditions, these factors don't matter?

This post has been edited by AndyH-ha: Apr 11 2007, 19:29
Go to the top of the page
+Quote Post

Posts in this topic
- AndyH-ha   potential mp3 difficulties   Apr 8 2007, 22:49
- - LANjackal   Pretty deep post there. I think the point of ABX t...   Apr 9 2007, 01:57
|- - Lyx   QUOTE (LANjackal @ Apr 9 2007, 02:57) Pre...   Apr 12 2007, 05:05
- - Light-Fire   QUOTE (AndyH-ha @ Apr 8 2007, 16:49)...   Apr 9 2007, 02:34
- - Silversight   QUOTE (AndyH-ha @ Apr 8 2007, 23:49)...   Apr 9 2007, 02:35
|- - kjoonlee   QUOTE (Silversight @ Apr 9 2007, 02:35) Q...   Apr 9 2007, 06:09
|- - pepoluan   QUOTE (kjoonlee @ Apr 9 2007, 12:09) I wa...   Apr 9 2007, 07:04
|- - Silversight   QUOTE (kjoonlee @ Apr 9 2007, 07:09) I wa...   Apr 9 2007, 23:40
- - AndyH-ha   Fatigue Fatigue and acclimation are common for ot...   Apr 9 2007, 06:04
|- - Light-Fire   QUOTE (AndyH-ha @ Apr 9 2007, 00:04)...   Apr 9 2007, 07:00
- - MichaelW   QUOTE (AndyH-ha @ Apr 8 2007, 15:49)...   Apr 9 2007, 06:25
- - AndyH-ha   Yes, if such fatigue is real, it may well effect a...   Apr 9 2007, 06:57
- - AndyH-ha   QUOTE You can quickly (seconds) switch between fil...   Apr 9 2007, 09:01
|- - Light-Fire   QUOTE (AndyH-ha @ Apr 9 2007, 03:01)...   Apr 10 2007, 01:42
- - Steve Grant   A very interesting subject. If you were to think ...   Apr 9 2007, 10:54
- - AndyH-ha   I did not challenge anything. I ask some questions...   Apr 10 2007, 04:44
- - Gabriel   All psychoacoustic-based lossy encoders are using ...   Apr 10 2007, 09:53
- - pdq   If your sound reproduction system deviates signifi...   Apr 10 2007, 13:59
- - Axon   Speakers regularly vary by 10-20db in their freque...   Apr 10 2007, 15:15
- - Axon   Bump. I'm not letting this one die. Is everybo...   Apr 11 2007, 16:38
- - AndyH-ha   QUOTE All psychoacoustic-based lossy encoders are ...   Apr 11 2007, 19:29
|- - Polouess   QUOTE (AndyH-ha @ Apr 11 2007, 20:29...   Apr 11 2007, 20:14
- - Axon   I guess the basic problem is that there's no c...   Apr 11 2007, 20:07
- - MichaelW   Has anyone done tests to see if two samples can be...   Apr 11 2007, 23:56
- - 2Bdecided   Last point first: Clearly a particular lossy encod...   Apr 15 2007, 04:49

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:


RSS Lo-Fi Version Time is now: 28th November 2015 - 16:58