Mp3, Aac, Wma?

[LDanix]

Hello everyone, from a new user.

I haven't seen much about this topic, so I thought I'd make it a thread.

I have recently purchased a Creative Zen Touch (harddrive MP3 player). It supports MP3 and WMA lossy compression formats (WMA lossless does not work). I am looking to place my CD collection on it; however, I do want to retain as much of the original sound quality as possible.

What experience have the experts had with MP3 and WMA at the highest bitrates and does a variable bitrate really make much difference? Which format do you believe is supperior for overall sound quality? Should I have purchased the iPod which supports AAC instead? Has anyone tried to rip SACD and/or DVD-A; if so, what software and which compression format were used?

 

[MDS]

MP3 is the oldest and therefore most widely supported of all the lossy compression ("perceptual coding") schemes followed closely by WMA.

Choosing a bitrate is a tradeoff between file size and sound quality - in general the higher the bitrate the higher the quality although it's difficult to make direct comparisons between different encoding formats at the same bitrate; ie WMA at a lower bitrate often sounds comparable to MP3 at a higher bitrate. Keep in mind that it is all subjective; I personally use 192 kbps MP3 and find it perfectly acceptable for nearly all my music. I have never been able to hear a difference between 192 kbps and anything higher so I stick with 192 kbps.

The idea behind variable bitrates is that the encoder will vary the number of bits used to encode each section of the music so that fewer bits are used to encode the simple parts (eg. you won't waste 192K bits to encode a second of silence as you would were you using constant bit rate). The downside is that very few players can correctly calculate the length of the song if a variable bit rate is used.

I would suggest ripping your cds and saving the waves on a large external hard drive (getting cheaper every day). That way you will always have the original and not have to re-rip the actual cd. You can then experiment with different encoders and bit rates to find what you like. That's what I do - I've ripped over 500 cds and have ~5000 waves and mp3s.

 

[kay]

I would also say go with 192 kbps MP3. For car/mobile use you will never know the difference even with 160 kbps.

 

[JoeE SP9]

I tried 192kbps and it was not suitable for home listening. 256kbps was acceptable to my audio buddies whose ears I trust. I felt this was a good trade off. It allowed me to get 3.5+ hours of music on a 700mb CD-R. It works well for parties and other social occasions.

 

[Rob Babcock]

I'd suggest you go no lower than 256 kbps, but be advised that the higher the bitrate you use the worse the battery life of your player will be. This is because the harddrive runs more as the disc will have to be accessed frequently to refill the memory buffer.

 

[LDanix]

Does it matter if I use MP3 or WMA? Can you tell any difference if the quality of the sound between 256 or 320 (or higher), perhaps. I have a set of Sony MDR-EX71SL earbuds that allow me to hear every little digital artifact. I'm trying to figure out what bitrate I can use that will minimize the audio junk and maximize battery life, also. I have been using WMA VBR which gave me an average bitrate around 400, it sounds great, but it sucks down the battery like there's no tomorrow.

 

[JoeE SP9]

I have not tried WMA VBR so I can't comment on that. As far as MP3 goes I can hear the difference between the original and an MP3 at 312 KBPS. For serious listening I prefer uncompressed audio. You will just have to balance battery life against sound quality. I would imagine that headphones/earphones would accentuate any little nasty resulting from the compression. I can't say for sure. I detest headphones and gave my Stax SR84's to a buddy.

 

[LDanix]

Can you recommend a high quality recording that I can use for my comparisions of the different formats/bitrates? Maybe I'm being a little too nutz about my portable audio experience but I'm looking for a CD/Track that may be considered to be of the highest possible quality to do my tests.

BTW, thanks for all your input!

 

[MDS]

First, avoid most CDs made since 1995 or newer as the vast majority are highly compressed and definitely not ideally mastered.

Names that often come up as being conservatively mastered and of very high quality are Steely Dan and Dire Straits. 'Audiophile' labels like Chesky and Telarc may be a good choice, if you like the kind of music that they offer.

Despite all the wrangling over bit rates and formats, it has been shown time and time again that for most rock/pop/dance/disco type music one will be highly unlikely to distinguish between an uncompressed wave and a sufficiently high bitrate mp3. However, there are a few types of music that don't encode particularly well and they are organ music and anything with very high frequency instruments like horns.

The song Vehicle by The Ides of March is about the only song in my collection where I can actually hear a difference between the uncompressed wave and the 192 kbps mp3. The song has a very shrill horn and is highly annoying, even on the wave. The mp3 accentuates the highs even more.
That's the kind of song to seek out if you really want to compare formats - for just about all other rock/pop there won't be much difference.

 

[JoeE SP9]

I would suggest using a couple of selections from your own collection that you think are recorded well. They should be ones that you are familiar with. In order to really determine what your minimum bit rate should be you need more than one selection. Try them back to back uncompressed versus compressed or minimum compression versus maximum compression. You will either reach a point where you can not hear compression or you will reach a point of acceptable trade offs. When you're happy with the results start ripping.

 

[LDanix]

Thanks for all the suggestions. I have blind-tested three selections vs the original uncompressed versions and have made my decision.

I started with an uncompressed WAV ripped using Exact Audio Copy and three compressed copies; two MP3s and one WMA. All were encoded with variable bit rates; however, the MP3s were bound to bitrates between 192 and 320; and 256 and 320. Both were encoded using the LAME 3.96.1 encoder. The WMA file was encoded at the highest quality Windows Media Player would allow (average was around 430 kbps).

The first file to be indentified as the worst reproduction in all three tests was the 192 kbps MP3. On all three, it only took a few seconds to identify the digital artifacts with the higher frequencies; I was mainly listening to the cymbals for these artifacts. The remaining MP3 and WMA were extremely close on all three. I identified the WMA as being the closer to the original for the first song; and the MP3 as the closer to the original for the remaining two songs. The MP3 had a "smoother" playback that more "true" to the original. On the second song, it took over 30 minutes of listening to the original then to the encoded versions (about 30 second clip) before I identified the difference between them. Once I figured it out, it was easier to identify on the other song.

Needless to say, MP3 VBR with a lower limit of 256kbps won 2 to 1. And since the sizes of the files are going to be much smaller, I'm not even going to second guess myself.

Thanks again, Everyone!

 

[JoeE SP9]

Your results pretty much mirror my own. I still find that for serious listening I prefer uncompressed files but for anything else MP3 at 256k works for me.

 

[LDanix]

Yeah, nothing can ever beat uncompressed for the best listening. It would be nice if these harddrive MP3 players were large enough to hold the uncompressed WAVs for enough CDs.
Just out of curiosity, have you tried any "Lossless" codecs? Do they really sound precisely like the original?

 

[WmAx]

Thanks for all the suggestions. I have blind-tested three selections vs the original uncompressed versions and have made my decision....

....The MP3 had a "smoother" playback that more "true" to the original. On the second song, it took over 30 minutes of listening to the original then to the encoded versions (about 30 second clip) before I identified the difference between them. Once I figured it out, it was easier to identify on the other song.

Do you care to share the samples used? Did you use special samples that could be considered codec busters? What methodology did you use for comparisons after encoding the files? Did you decode them to PCM again before comparing? Did you time synch the files in an audio editor before the comparison? Did you analyse the spectrum data in the files? Did you find it difficult/fatiguing to compare a single 30 second sample in a DBT for 30 straight minutes? What were the total test scores?

-Chris

 

[JoeE SP9]

Just out of curiosity, have you tried any "Lossless" codecs? Do they really sound precisely like the original?

I have been intending to set something up with my informal listening group. I will be using FLAC to start with. If you have any suggestions about other "lossless" compressors please let me know, although too many CODECS at one time make testing overly long and complicated.

 

[LDanix]

Do you care to share the samples used? Did you use special samples that could be considered codec busters? What methodology did you use for comparisons after encoding the files? Did you decode them to PCM again before comparing? Did you time synch the files in an audio editor before the comparison? Did you analyse the spectrum data in the files? Did you find it difficult/fatiguing to compare a single 30 second sample in a DBT for 30 straight minutes? What were the total test scores?

-Chris

Chris- I didn't go into nearly that much detail. I only used my ears. I figure that if I can't hear an artifact or a difference, then there's no need to worry about it.
I listened to all samples through my Creative Zen Touch with my Sony MDR-EX71SL earbuds, since these were what are going to ultimately be used for my listening.
I kept the MP3s and WMA in their formats. I did not decode them back to PCM.
Yes, it was fatiguing to listen to 30 second clips over and over again. I rested for a while between each set of clips.
I don't have anything that is from 2005 so I used some fairly new recordings that I have felt sounded better than the rest in my collection. Here's what I used:

Artist, Album, Song
Avalon, The Creed, All
Avalon, The Creed, The Good Way
Rachael Lampa, Live for You, Day of Freedom

I'm not familiar with "codec busters". I gather from their name that they are extremely difficult to encode accurately. ? What can you tell me about them?

 

[LDanix]

I have been intending to set something up with my informal listening group. I will be using FLAC to start with. If you have any suggestions about other "lossless" compressors please let me know, although too many CODECS at one time make testing overly long and complicated.

Joe- The only lossless codecs that I am familiar with are ADPCM and WMA Lossless. I have heard of FLAC a couple times but never really researched it much.

I'm more interested in WMA Lossless because any device that carries Microsoft's "Plays For Sure" seal is supposed to play these. I am going to eventually set up a media server in my home which will have several "Plays For Sure" devices. All will be wireless, so there won't be enough bandwidth for uncompressed PCM or WAV data.

Knowing that I won't have to give up audio quality for the convenience of the media server would just make the feel all tingly inside.

 

[MDS]

Setting up a listening test to compare lossless codecs is worthless. If the codec is truly 'lossless' it will decode exactly to the original uncompressed PCM.

Such a test would be like zipping a text file with both pkzip and winzip then unzipping them to see if they both result in the original text file.

 

[LDanix]

Setting up a listening test to compare lossless codecs is worthless. If the codec is truly 'lossless' it will decode exactly to the original uncompressed PCM.

Such a test would be like zipping a text file with both pkzip and winzip then unzipping them to see if they both result in the original text file.

You could be right. However, there are no tests that I have found that have determined that these codecs are truly "lossless" (if you can find one, please pass it on). Everyone here knows that you can't always believe hype or what the companies claim about their products. That's why there should be an independent test. If all the codecs sound precisely like the original, then great; what's left to determine is which codec can compress the PCM to a smaller package than the others.

 

[WmAx]

...Chris- I didn't go into nearly that much detail...

Then your test could have likely introduced false positive results for difference(s) that may not necessarily be due to inherant limitations of the file format(s) or compression level(s) used.

-Chris