MP3Pro??

[Justin D]

Has anyone ever heard of this format? It claims to be higher quality than regular MP3's but is only 64kbps. Does anyone have any experience with them?

Thanks.

 

[Unregistered]

It is similar to MP3 but is supposedly optimized for lower bitrates. The claim is that MP3 Pro at 64kbps is similar in quality to MP3 at 128kbps. Given that 128kbps MP3 rarely sounds good to me, I was not impressed with MP3 Pro.

Probably works well for speech or low quality streaming though.

 

[Justin D]

Ahh, I see.

So, what format does everyone recommend for ripping CD's? I know uncompressed formats are best but that takes a lot of space.

So what say you?

 

[Rob Babcock]

This probably won't be very helpful, but I haven't ever been very satisfied with MP3, either. Rock music is borderline acceptable with 320 kbps rates or the very best VBR, but that's not good enough for classical music, IMOHO. My fling with MP3 was brief- I tried the DMX pluggins for MMJ and many encoding rates, and I could never really get acceptable sound (by my standards).

 

[Unregistered]

I didn't mean to imply that mp3 in general sounds lousy to me, just that 128kbps is definitely not 'cd quality' as it used to be billed. I have been able to hear the difference between 160kbps and 192kbps mp3, but not between 192kbps and anything higher. Hence, I chose 192kbps for my mp3s.

MP3 is the best choice if you want universal compatibility as all devices support it. WMA is rapidly catching up with many devices supporting it. AAC sounds very good at lower bitrates than a comparable MP3 but is the least supported of all popular compression formats. Pick your poison.

 

[Rÿche 1]

My 1,000+ mp3's are encoded with Lame VBR bitrate. VBR rules, because instead of a constant bitrate, it will give you the variable instead. I use 32-320kbps. Perfect sound. Near cd quality. If you use a constant, say 192kbps, not all parts of the song will need the 192. Also, some parts of the song need more than 192. Hope I explained that alright.

 

[Unregistered]

You explained it correctly but there is one additional consideration which may or may not be relevant to any one particular person. Many players cannot correctly determine the length of the file if it is VBR. Lame has their own proprietary info in the header to help with that but not many players, that I know of, actually know how to read it and calculate the correct length.

Winamp does, but Windows Media Player does not. Those are the only two I use regularly.

 

[Rob Babcock]

VBR is the way to go if your playback hardware will support it. VBR uses the highest bitrates for the most complex passages, scaling it back where a low bitrate will suffice. The result is a better sounding track & a smaller file.

Most of the CD & DVD players I own will play back a VBR but the time indicator is all screwy.

No matter how high the bitrate, though, I don't think it sounds like the original. Maybe for casual listening, but the music is more fatigueing over time, and acoustic stuff particularly just doesn't sound natural. The sound is also a little edgey. But I'm no expert on codecs, and maybe there's a lot better encoders out there that I haven't tried yet. Ultimately I don't need codecs, either. For me it just something to mess around with.

 

[WmAx]

I believe most of the adversion to lossy compression schemes is due to psychological bias. Standard mp3 format(which actually has a broad selection of encoders available) seems to be one of the most ill-regarded. Perputuated largely by the poor quality encoders and poor quality files available from most sources such as Kazaa, etc.?

The listening test forum over at hydrogenaudio.org may be a very good read for those interested in audibile quality/differences in lossy compression schemes.

In objective ABX tests, some people have troulbe discerning bitrates over 128kbps from the reference. I can't point to any 192kbps examples, but people reporting personal ABX trials at this rate generally have a tough time, exceptions being some extraordinary file samples that hilite the artifacts and/or people with exceptional sensitivity to artifacts compared to other test subjects. So far a 192kbps or higher formal test has not been produced that I am aware due to the reluctance of most people to take part in the trials.(the smaller the differences, the longer and more fatiguing the test).

I don't think the poor reputation of mp3 is entirely deserved.

I recently started using Lame 3.96 MP3s encoded at 256kbps for my portable hard drive based audio player. I previously hauled around a case of CDs and a discman. I'm pretty picky about sound, and use a highly regarded headphone for my portable audio(Sony MDR-CD3000); I don't notice any general difference from CDs in casual listening. I should point out that I have not performed a DBT or ABX test on the MP3s vs. source files. If their is any audible difference(s) in parts of the music, they seem to be subtle to the point that I would not know it unless I had the original to directly A/B with the sample. So, it hard for me to justify using CDs over a proven good quality MP3 encoder such as Lame at a sufficient bitrate. However, it is undersandable that some people may not be able to get over the psychological bias against the format. IN that case, it does not actualy matter if the file is transparent in reality or not. Just my personal views on this matter.

-Chris

VBR is the way to go if your playback hardware will support it. VBR uses the highest bitrates for the most complex passages, scaling it back where a low bitrate will suffice. The result is a better sounding track & a smaller file.

Most of the CD & DVD players I own will play back a VBR but the time indicator is all screwy.

No matter how high the bitrate, though, I don't think it sounds like the original. Maybe for casual listening, but the music is more fatigueing over time, and acoustic stuff particularly just doesn't sound natural. The sound is also a little edgey. But I'm no expert on codecs, and maybe there's a lot better encoders out there that I haven't tried yet. Ultimately I don't need codecs, either. For me it just something to mess around with.

 

[Rob Babcock]

You're right about DBT being very difficult. I'm a believer in DBT, provided the associated gear is good enough. But I will also concede that occasionally something you can't immediately ID in an ABX comparator-type test can still lead to fatigue over time. I don't have a good explanation for it, but I've just experienced it too often myself.

While it's not super scientific, I have conducted my own tests using some willing guinea pigs. I've synched identical CD players up, one with a compressed disc and the other an identical compact disc (same track sequence & level matched as closely as I could) with various bitrates. Even at 256 kbps the subjects in my little test didn't have a lot of trouble hearing the diff.

Of course, that's really only an indictment of Nero & MusicMatch Jukebox encoding in conjunction with the gear I was using. There may be a lot better encoders- if you know of them, I certainly wouldn't contradict you. MP3 may have a bit of an undeserved bad rep, but in my experiences with it, I've never heard anything that approached "hi fi."

But I also think, after listening to a lot of DVD-A (and to a lesser degree, SACD) that Redbook itself is only marginal. Even the best MP3 is only going to approximate Redbook, which is noticably inferior to good 24/96.

Comparing MP3 to DVD-A is like comparing a covered wagon to an F-14, IMOHO.

 

[Rob Babcock]

BTW, I've never used an MP3 that was downloaded, nor have I ever heard one. My only experience with the format has been ripping from my own compact discs. I've used only Nero & Music Match Jukebox, with bitrates of 128 up to the highest rates the software would support. I did try lower rates for the hell of it, but none of my machines will recognize really low bitrates. And frankly, 128 sounded gruesome enough to discourage going any lower.

One further observation- certain types of material encode pretty successfully and some things don't. My experience has shown the two things that really don't work very well when compressed are massed strings and highly distorted rock guitar. The reading I've done confirms these things give most codecs fits. For plain-jane pop, even MP3 > 256 kbps is not terrible. But even 320 kbps doesn't sound very good with heavy metal or classical, IMOHO.

 

[WmAx]

But I will also concede that occasionally something you can't immediately ID in an ABX comparator-type test can still lead to fatigue over time. I don't have a good explanation for it, but I've just experienced it too often myself.

SPeculation is okay, as long as you realize that's what it is. However, ABX is one of teh most sensitive of comparator tests, far more so than casual comparisons.

While it's not super scientific, I have conducted my own tests using some willing guinea pigs. I've synched identical CD players up, one with a compressed disc and the other an identical compact disc (same track sequence & level matched as closely as I could) with various bitrates. Even at 256 kbps the subjects in my little test didn't have a lot of trouble hearing the diff.

Their is a program called PCABX, i recommend you use this and perform ABX testing or use Foobar with the ABX component option. I suspect something was in error with your testing methodology if you 'no difficulty' picking the 256kbps samples. I believe both MusicMatch and Nero use Fraunhofer developed codecs. This is supposedly one of the best options for encoding. At the very least, it should very difficult to score well in a blind test with most music at bitrates this high.

I highly recommend reading the Listenig Test forum at hydrogenaudio.org to get a good idea of how to set up a proper test. To get a feel, take one of the past tests, most of the test files and instrcutions are still availble in the links listed in the posts.

There may be a lot better encoders- if you know of them, I certainly wouldn't contradict you.

Lame has scored well in ABX tests. Another option.

But I also think, after listening to a lot of DVD-A (and to a lesser degree, SACD) that Redbook itself is only marginal. Even the best MP3 is only going to approximate Redbook, which is noticably inferior to good 24/96.

I would caution such an absolute statement. Their are no auditory tests that have held up under scrutiny of peer review and confirmed an audible improvement of hi-rez formats for playback. Actually, too the best of my knowledge, RBCD exceeds the parameters that have been established for playback transparency in perceptual testing.

-Chris

 

[WmAx]

One further observation- certain types of material encode pretty successfully and some things don't. My experience has shown the two things that really don't work very well when compressed are massed strings and highly distorted rock guitar. The reading I've done confirms these things give most codecs fits. For plain-jane pop, even MP3 > 256 kbps is not terrible. But even 320 kbps doesn't sound very good with heavy metal or classical, IMOHO

If you would do this, rip some songs from CDs of what you consider to be 'problem' samples. Go into an audio editor and isolate a 3 to 5 second sample for each 'problem area'. Save thiese wave samples. I would like to have these to encode/experiment and ABX. If it's too much trouble, that's understandle.

-Chris

 

[Unregistered]

I can easily create a 3 to 5 second sample for you to a/b. The only problem is I don't know exactly what types of music give the average encoder a problem.

I've done some of my own a/b testing when trying to decide on what bitrate to use. My collection is mostly rock/pop/disco/funk music. I could discern a subtle difference between 160kbps and 192 kbps on one particular rock song with alot of acoustic guitar, but have never been able to reliably tell the difference between 192 kbps and anything higher - so I chose to use 192 CBR.

 

[WmAx]

Nice of you to offer. However, you stated yourself that you believe that at higher bitrates MP3 is effective for your purposes. I wanted to examine samples of which that other user was having problems/complaints.

If you want to test/push the encoder: credible people on hydrogen audio have said that harpsichord solos make for easy detection of compression artifacts compared to most other musical instruments in ABX tests. Classical music in general is regarded as being harder to encode effictively at low to moderate bitrates.


-Chris

I can easily create a 3 to 5 second sample for you to a/b. The only problem is I don't know exactly what types of music give the average encoder a problem.

I've done some of my own a/b testing when trying to decide on what bitrate to use. My collection is mostly rock/pop/disco/funk music. I could discern a subtle difference between 160kbps and 192 kbps on one particular rock song with alot of acoustic guitar, but have never been able to reliably tell the difference between 192 kbps and anything higher - so I chose to use 192 CBR.

 

[Unregistered]

I did some ab (not blind) testing on a good hifi using a cd composed of alternate original-compressed pairs of samples. I could only barely tell the difference, and it was hard to hear, on 128k cbr compression.

But there may be something about the way the data is reconstructed too. I compressed, then decompressed, and burned all files as wav data onto a cd. That certainly would have resulted in a higher quality output than playing from a portable mp3 player or something. But the actual loss (some kind of fraunhofer encoder) was minimal, even at 128. I was surprised, thinking you'd need at least 256k to get so close to the original. Even 96 was ok, but becoming obvious, and unsurprisingly, 64k sounded pretty bad.

 

[WmAx]

I compressed, then decompressed, and burned all files as wav data onto a cd. That certainly would have resulted in a higher quality output than playing from a portable mp3 player or something.

I am not aware of any credible studies demonstrating portable music players to be audibly inferior to home component players. Do you have reference to such a study?

-Chris

 

[Westrock2000]

Has anyone ever heard of this format? It claims to be higher quality than regular MP3's but is only 64kbps. Does anyone have any experience with them?

Thanks.

I'm sorry did you say something?


I have heard from several places, but not actually red the spec, that MP3pro deletes most if not all the information above 10 Khz, and then "sythensis" it back upon playback. So in effect it gets rid of half of the audio spectrum. This may be how it gets the smaller files size, by focusing on the meat of the music and disregarding what most cheap components can't playback. Add to it more efficient psychoacoustic modeling and stuff.

I have yet to really read any regards on lossless schemes like AAC lossless, FLAC, or Monkey. Although there compatibilty is severly lacking compared to MP3....although for archiving it wouldn't matter as much. I think I would use WMA if they would offer something more than 192kbps....that just isn't high enough to make me comfortable, which may just be pyschological. But I do 320kbs CBR LAME on CDex. I usually archive my real CD's and then burn copies that can get scratched and beat up.