When is MP3 loss detectable?

fwrunner2017

New member
I'm not sure if this belongs here, or in the Computers/Digital recording forum, so please move it if it does not belong here:

I am a non-professional, but enjoy listening to my CD's and other purchased music (mostly iTunes) on my PC and my Android phone. I have always used the mp3 format, and over the years, have used some higher bit rates as they became available. Still, I like to keep file size down to manageable size, since my phone has only 32 Gigs of storage on the SD card.

Mainly, I rip my CD's to .wav format and keep the songs on my hard drive. For the iTunes songs, I burn the protected files to CD, then rip as unprotected .wav files.
With the wav files, I like to add limiting to make the music loud without clipping. I am using Audacity on a Windows 7 computer.
Once I have the sound I want, I export it to .mp3 format, with settings: Preset Standard 170-210 Kbps, Variable speed Fast, and channel mode to Joint Stereo.

Using the onboard sound system on my PC and a set of Sennheiser HD-497 phones (I have other, cheaper ones but they all produce the same result) I am unable to tell the difference between the original .wav file and the compressed .mp3 file.

I tried an experiment, creating successive generations of the same .mp3 file by encoding and saving, closing the file, re-opening, and repeating the encoding-saving process for up to 10 generations. So I have an mp3 of an mp3 of an mp3...
I started out with an uncompressed wav file. The songs were rock, R.E.M., Bryan Adams, and Green Day, all of which had high volume levels.

When I reach the 3rd generation of the mp3, I can hear a very slight difference. There appears to be a "swishy" sound on the high notes, especially with symbols.
At the 5th generation, the distortion becomes easily detectable, and at 10th generation (which I know is ridiculous), the sound is pretty much unbearable.

So, I'm reading up on the caveats of the mp3 format, and that it doesn't take a trained pro to detect the distortion on even a 1st generation mp3 encoded at even the highest bit rates using pretty much any encoder/decoder.
BTW: Audacity is using the Lame encoder. I don't know what iTunes uses, but it all sounds the same to my ears.

I am starting to wonder why I can't detect any difference between the original wav file and the mp3 until I've done three generations of encoding/decoding. I start to think that it is my computer's sound system that is the weak link, and not the mp3 encoding process I am using.
My system is an MSI P67A board with Realtek "High Definition" audio. I do not have a separate sound card.
I am listening to the music directly from the analog output of the computer using headphones.

Next I put the original CD from which I had ripped the songs into a Sony 5-CD changer and connected its analog output to my Onkyo receiver. These components are nothing special, but still I would say that they're better than the Realtek sound "card" on my computer.
Sure enough, there is a significant improvement in the playback I get directly off the CD on the home system as compared to the computer playing either the CD directly through the Realtek sound or from the ripped .wav or mp3 files on the hard drive.
The stand-alone sound system produces much crisper tones, sharper drum raps and all of that.

This leaves me wondering what I need to do if I want to have the same sound quality on my computer (forget about the phone - I realize I'm not going to get close on that device) as I do on the home stereo system.
I have no doubts that I will get cleaner sound from a dedicated sound card than I do from the built-in Realtek. I checked out one blog and found one card that comes highly recommended is the Asus Essence STX II. That card is selling for close to $200 on Amazon, but there are some not-so-good reviews as well.
I'm not stuck on any particular brand, so I am looking for someone to point me in the right direction for either a new or used card.
I don't want to spend more than $200 for sure, and would love to get something considerably better than the Realtek for around $100.
Considering that my best set of headphones are the Sennheiser HD-497, with other less expensive phones available to me, I don't want to spend more on the sound card than is appropriate considering my headphones. That said, I would not rule out upgrading the headphones as well - especially if I could get a good used pair.

Oh, and I forgot to mention that I also have a set of Cervin Vega speakers that I rarely use, but are in perfect working condition. Thing is, I can't crank the audio in the house, so I listen only through headphones, while the speakers are beside my bed for playing "white noise" and other sleep inducing audio

I'm not expecting a silver bullet that will cost me under $200, but hopefully, I can make one or two reasonable investments and get a more enjoyable sound experience.
My main listening genre is Rock (not too much metal - mostly loud fast rock songs from R.E.M., Bryan Adams, Green Day, etc). But that is not to say I don't listen to real Classical music (Mozart, Beethoven, etc) when the mood strikes.

Thanks for your input.

FW
 
I can hear the difference immediately. 320kb constant rate mp3s are okay but still noticeable.
TBH, I didn't read your entire post because it falls under "TLDNR", but in short, if you want great quality digital files get a backup hard drive and encode as Wav, Flac, or Alac.
 
Cool experiment. I don't hear a difference either. Speaking for myself, my ears just aren't that sensitive to pick up minute frequency loss. I'll notice if something annoying has been added, especially to the midrange, which happens frequently with home mastering. But if it's just loss, it has to get a more pronounced to get my attention. Then again my ears are 50 something.
 
It does depend, quite a bit, on the encoder. LAME 3 is popular and pretty good. There are a range of them out there though. Playback machine matters almost as much as does the final conversion to air.
Personally I like to use .wav or .FLAC for my D.A.P. but if I haven't an option I'll use MP3 320.
I don't hear some of the issues because I have tinnitus and changes in the related area are masked anyway but above & below I can often hear the loss but most certainly hear the artifacts added ( swish, tunnel tone etc) from too low a com/reduction rate or poor encoding. I usually play digital files, other than CD, on a Colorfly C3 which has wonderful sound, (though no fun features like shuffle), and have found that on that player, as opposed to the other dirt cheap and semi expensive players I've used, doesn't hide the bad encoding. Sadly poor MP3 audio has become the norm. I use fLAC 8 because I don't really care how long it takes to encode from a CD or .wav file and only have about 32G on the DAP so the, probably, minute difference in file size is useful and since the decoding time is constant there's no issue for me. I do believe FLAC5 is quite popular.
 
If you're just wanting a high quality monitoring chain, then yeah a DAC with a headphone output is probably the way to go. If you also want a high quality input path for recording audio, then an audio interface would be the way to go.
 
Each coding introduces errors. Your ears and brain set the point at which you can detect it, and of course the music itself can help or hinder.

The mangling is progressive. Often when editing I can see glitches in the waveform, perhaps from a sloppy edit, but sometimes I cannot hear them - but they are there.

When you improve your monitoring you suddenly hear things you had not heard before. When you buy a better microphone, you hear the same - very small differences. The untrained person is oblivious. Does that Stradivarius sound better than the cheaper one? To the player certainly - how about to others, less violin minded?

How many of us getting older, know that our hearing is losing the top end, yet can still hear what we remember hearing? Am I, in my 50s, less able to discriminate? On paper, certainly - but in real life?

Those ridiculous audiophiles who write the silly hi-fi reviews, detecting sonic differences in that £1000 IEC cable they just bought - what exactly do they think they are hearing? Is this the same as the difference between 320K and 310K mp3s (if such a small difference was possible)?
 
If you're just wanting a high quality monitoring chain, then yeah a DAC with a headphone output is probably the way to go. If you also want a high quality input path for recording audio, then an audio interface would be the way to go.

He's still stuck with the cheap computer optical device. The nicer interfaces and cards can pair up well with a decent home optical player.

Some used $20 player can kick ass;
"Best Buys of the Year
1. Sony DVP-PR50P and its twin DVP-SR200P (previously reviewed here). These ridiculously cheap players surprisingly play CDs really well. The DVP-PR50P is no longer available, so grab the SR-200P while you still can. Some people use these as transports, but I think they are missing the point. The analogue outputs are the real stars here, an uncanny mixture of detail and musicality that is missed by the majority of expensive digital products. If the analogue outs don't sound good in your system, question the balance of your system first".
 
Thanks guys; You gave me a lot to think about. I have been planning to use .wav files on my pc (I do have the hard drive space), and only encode to mp3 for the phone, where it really doesn't matter anyway. I can use .ogg for the phone as well, but since I can't tell the difference on that device, I will use a higher compression ratio so to fit more music on the phone, which has a max 32Gig micro SD card.
I suppose that I will someday upgrade my sound system. The USB DAC sounds like my best bet, since I'm not interested in recording. Nice thing about the USB DAC is that I can use it on any computer - like a tablet or notebook.

I ran a rudimentary hearing test on myself yesterday. I can hear up to 15Khz @ 0.8db (not calibrated; that was the level I generated the tones at with Audacity) using the Sennheiser phones, but cannot hear that sound from the earbuds - I guess they just can't put out enough power at that frequency for me to hear it.
I am also 50+, so my hearing is not quite what it used to be, when I could hear the flyback oscillator (15750Hz) in my TV set. If I can still hear up to 15K, then I guess my hearing is still pretty good. I have always had a sensitive ear; I think my computer's sound system is what is the most limiting factor.

Sorry about the "TLDNR". I do tend to get long.

FW
 
I turned 50 in 2000 and my right ear was almost useless by then. I was hot on getting a Aune DAC (as somewhat affordable), but who am I kidding : ) I found a import DAC that does all the DSD and PCM for $80 and that is good enough. It isn't for headphones. hahah It gives me line out on Mini, so I can use my big stereo.

The half decent DAC should probably have something like 600-Ohm Beyer DT880 fones. You mentioned used. If you can play the waiting game, some old 600-Ohm AKG-240 would work
 
Wikipedia: "At a bit rate of 256 kbps, the quality hardly differs from a CD."
This "to hear or not to hear" is an endless discussion within audiophiles (=not files :p for those who intend to read this wrong)

Why don't you rip in Flac? Thats equal to wav, but lossless compression. And also possible without compression.
Don't understand eitherway that flac isn't that popular yet. I love it and use it all the time. :listeningmusic: When recording 32 bit floating, 96khz, non compression, and if exporting then with some compression. Higher quality within a smaller file (that's with f, not with ph LOL )
 
I'm not sure if this belongs here, or in the Computers/Digital recording forum, so please move it if it does not belong here:
I moved it over to the Digital Recording forum.

I'm like RayC, tinnitus keeps me from hearing any real difference in quality of MP3s. Still, I use 320 whenever possible. I will lower the data rate down so a file will be under the size limit of the upload requirements of the forum. 10MB, I believe it is.
 
I use LAME, BLADE, and GOGO and am usually disappointed on the first run with mp3. I may post a 160k and say this sucks, but the tone balance is right : ) My 2gb player went dead, but I'd was stuffing it with .wav or, FLAC files, anyway. A lot of what I convert (old tapes, etc.) to mp3 will be 22k and some will get some fake mastering at high rates.

I find the free TASCAM Hi-Res Editor really nice;
Product: TASCAM Hi-Res Editor | TASCAM
 
My tinnitus keeps me from hearing the difference on first generation done with a good LAME encoder, but try uploading an MP3 to soundcloud (who mangle it with their own encoder) and I can sure hear it.
IF you're doing any recording, I guess you don't need a true audio interface, but it might be your best bet anyway to get decent D-A conversion. The Steinberg UR22mkII just had a price drop at Sweetwater - $130.
 
320 is pretty acceptable...almost negligent MP3 conversion "hash".
Less than that it gets more and more noticeable..but you have to know what to listen for.

It's not really like a hearing test, where you just focus on your ability to hear the very high end. It's more about the artifacts that occur on transients and points of transitional audio "blending"...kinda like what you see on digital TV when there's a fade to black, you get the artifact halos...etc..

Start with a really low quality MP3 conversion, so that the artifacts are obvious. Then focus on that sound VS the WAV, and then move up in MP3 quality and keep focusing on the artifacts. You just have to "zoom in" on it and be aware of what it is you're listening to.

The flip side...if you can't hear the conversion "hash"...don't worry about it. :laughings: ;)
 
320 is pretty acceptable...almost negligent MP3 conversion "hash".
Less than that it gets more and more noticeable..but you have to know what to listen for.

It's not really like a hearing test, where you just focus on your ability to hear the very high end. It's more about the artifacts that occur on transients and points of transitional audio "blending"...kinda like what you see on digital TV when there's a fade to black, you get the artifact halos...etc..

Start with a really low quality MP3 conversion, so that the artifacts are obvious. Then focus on that sound VS the WAV, and then move up in MP3 quality and keep focusing on the artifacts. You just have to "zoom in" on it and be aware of what it is you're listening to.

The flip side...if you can't hear the conversion "hash"...don't worry about it. :laughings: ;)

Gonna have to agree with this.

I hear the difference between wav and mp3. But if I didn't, then I surely wouldn't be doing what I do...
 
Also, the poster mentioned he was using variable rate, which I don't think translates music very well. hah
 
Yeah...I saw that and was going to comment...then I forgot about it.

Variable rates are kinda OK if the primary focus is steady streaming at any cost...but it's not the best option for maintaining stream quality.

I would always rather wait/download than worry about steady low-grade stream...though that doesn't bother a lot of people these days, since they've already opted for the lowest common denominators = convenience and instant gratification. :p
 
If an interface is on the cards (Boom! Boom!) the Zoom UAC-2 is said to have the best converter quality of any AI near its price. Still very good are offerings by Steinberg's UR22 and the Focusrites. If you wanted a digital input to D/A CDs the Native Instrument Komplete Audio 6 would serve you very well and its 4 analogue inputs mean it could be a fixed 'collection point' for two stereo sources.

But even the sub £100 Alesis i02 Express will beat Realtek. Heck! The 25quid Behringer UCA 202 does that!

You might also consider investing in a good headphone amplifier? The Behringer HA400 is not at all bad but £100 buys you better.

Dave.
 
Back
Top