Page 2 of 3 FirstFirst 123 LastLast
Results 16 to 30 of 31

Thread: MP3s sound like crap through speakers

  1. #16
    Senior Member bigshot's Avatar
    Join Date
    Nov 2011
    Location
    Hollywood U.S.A.
    Posts
    5,866
    Post Thanks / Like
    Blog Entries
    1

    Default

    I've already prepared my test. It took some time. It includes choral and orchestral music that is very well recorded and has a wide dynamic range and response. The clips were selected by an audiophile in this group as being difficult to compress. If that's OK, I'd be happy to administer the test to you. If it isn't, I would encourage you to set up your own controlled listening tests. I can help you and give advice on the proper way to do it. Controlled testing is how you determine exactly where your own line of transparency lies. If you want to do that, gather up a clip of around three minutes in length as a WAV or AIF file and I'll help you set up test files.
    CD Sound Is All You Need: http://xiph.org/~xiphmont/demo/neil-young.html
    AES Audio Myths Seminar: http://youtu.be/BYTlN6wjcvQ
    AES Damn Lies Seminar: http://youtu.be/Zvireu2SGZM

  2. #17
    Senior Member Larkenfield's Avatar
    Join Date
    Jun 2017
    Location
    Sedona
    Posts
    3,862
    Post Thanks / Like
    Blog Entries
    4

    Cool

    Set up a blind test on preamps for people who like to set up blind tests and watch them flunk because they forgot to listen.
    Last edited by Larkenfield; Jun-14-2019 at 04:13.
    "That's all Folks!"

  3. #18
    Senior Member bigshot's Avatar
    Join Date
    Nov 2011
    Location
    Hollywood U.S.A.
    Posts
    5,866
    Post Thanks / Like
    Blog Entries
    1

    Default

    I test every piece of equipment I buy. It's a good idea to do that because after the 30 day return window closes, you're stuck with it.
    CD Sound Is All You Need: http://xiph.org/~xiphmont/demo/neil-young.html
    AES Audio Myths Seminar: http://youtu.be/BYTlN6wjcvQ
    AES Damn Lies Seminar: http://youtu.be/Zvireu2SGZM

  4. #19
    Senior Member millionrainbows's Avatar
    Join Date
    Jun 2012
    Location
    USA
    Posts
    12,568
    Post Thanks / Like
    Blog Entries
    137

    Default

    The choice of sound sources, and their nature, would make a great deal of difference in a comparison test. For example, in the EMI Buzzcocks remasters, you do not want to listen to the distorted electric guitars for your cues. Rather, listen to "impulse" sounds, like drums, to hear differences. On the newer EMI Buzzcocks remasters, you can hear more "room" ambience on the drums. You could call it "air." Also, the vocals are a good indicator.

    With your proposed listening test, a choir is not a good choice, because there are no fast transients; and that's where jitter is more audible. it's all smoothed-out. Same with an orchestra. What you need is a variety of sounds.
    "The way out is through the door. Why is it that no one will use this method?"
    -Confucious

    "In Spring! In the creation of art it must be as it is in Spring!" -Arnold Schoenberg

    "We only become what we are by the radical and deep-seated refusal of that which others have made us." -Jean-Paul Sartre

    "I don't mind dying, as long as I can still breathe." ---Me

  5. #20
    Senior Member bigshot's Avatar
    Join Date
    Nov 2011
    Location
    Hollywood U.S.A.
    Posts
    5,866
    Post Thanks / Like
    Blog Entries
    1

    Default

    It took me a whole weekend to prep all the files for the test I assembled. I let the person who wanted to take the test first pick the tracks. You're free to construct a test using your own tracks. It won't change the results.

    The reason a choir was picked was because massed sounds like that are difficult to encode without artifacting at lower rates. Applause and choral sounds are notorious for creating problems. Transients aren't generally a problem. It's complex sustained sound.

    Jitter has nothing to do with compression.
    Last edited by bigshot; Jun-14-2019 at 16:23.
    CD Sound Is All You Need: http://xiph.org/~xiphmont/demo/neil-young.html
    AES Audio Myths Seminar: http://youtu.be/BYTlN6wjcvQ
    AES Damn Lies Seminar: http://youtu.be/Zvireu2SGZM

  6. #21
    Senior Member
    Join Date
    Jul 2013
    Posts
    1,948
    Post Thanks / Like

    Default

    Quote Originally Posted by bigshot View Post
    The reason a choir was picked was because massed sounds like that are difficult to encode without artifacting at lower rates. Applause and choral sounds are notorious for creating problems. Transients aren't generally a problem. It's complex sustained sound.
    Is a long tone of trumpet also difficult to encode at low rates like 128kbps? When I started collecting mp3s, I found popping noise in sustained trumpet sounds. A long time ago I purchased a mp3 of Takemitsu's Paths for solo trumpet by Hardenberger, whose bit rate was very low, 107kbps VBR. There are noticeable cracking/popping noises when a long tone is being played. I once ripped Miles Davis & Gil Evans CDs at a low rate (I think it was 128kbps or so) and found similar noises when Mile plays a long tone on trumpet. I have not heard anything like that with mp3s at 256kbps or higher rates.

  7. #22
    Senior Member bigshot's Avatar
    Join Date
    Nov 2011
    Location
    Hollywood U.S.A.
    Posts
    5,866
    Post Thanks / Like
    Blog Entries
    1

    Default

    128 is going to definitely have audible artifacting. But popping sounds in loud parts may also be distortion due to clipping. I've found some encoders alter the volume level a little bit, and if a track is normalized all the way up to the maximum peak, it can boost into clipping when you encode it. If you have a track that does this, try opening it in an audio editor and lower the overall volume a few percent. Then try encoding it. That might fix it. Of course upping the data rate to 192 or above would help too.
    CD Sound Is All You Need: http://xiph.org/~xiphmont/demo/neil-young.html
    AES Audio Myths Seminar: http://youtu.be/BYTlN6wjcvQ
    AES Damn Lies Seminar: http://youtu.be/Zvireu2SGZM

  8. #23
    Senior Member
    Join Date
    Jul 2013
    Posts
    1,948
    Post Thanks / Like

    Default

    Here is the waveform opened with Audacity. The trumpet is playing a long note here. The upper one is the bad mp3 (107kbps VBR), the lower one is the audio from youtube, which does not have the noise. The bad one apparently has steep changes of the amplitude with a period of about 41msec. I am wondering if this is due to the sound which is difficult to encode at a low rate or just a bad encoder software.

    paths_audio.jpg

  9. #24
    Senior Member bigshot's Avatar
    Join Date
    Nov 2011
    Location
    Hollywood U.S.A.
    Posts
    5,866
    Post Thanks / Like
    Blog Entries
    1

    Default

    That doesn't look like clipping. I guess you'll need to up the data rate a bit. 107 is a very low setting for stereo.
    Last edited by bigshot; Jun-16-2019 at 09:02.
    CD Sound Is All You Need: http://xiph.org/~xiphmont/demo/neil-young.html
    AES Audio Myths Seminar: http://youtu.be/BYTlN6wjcvQ
    AES Damn Lies Seminar: http://youtu.be/Zvireu2SGZM

  10. #25
    Senior Member
    Join Date
    May 2016
    Location
    NZ
    Posts
    1,128
    Post Thanks / Like

    Default

    A very interesting thread and just as a laymans observation, one thing that I have noticed is that with a low bit rate say 128 - 256 kbps played at volume on a decent set up through good equipment is that of distortion, this may not be heard through a PC.

  11. #26
    Senior Member
    Join Date
    Sep 2012
    Posts
    1,118
    Post Thanks / Like

    Default

    I discovered long ago that I definitely cannot tell the difference between WAV 16 bit and MP3 lame at 320 kbps. For some types of music I can go down as low as 192 kbps and still not detect any difference.

    The better the equipment used, and the younger the person being tested, the more likely it is that small differences may be discerned in very difficult pieces of music at MP3 192 kbps. Even so, I still doubt that at MP3 256 vbr or 320 cbr kbps there would be any significant results compared with WAV. With AAC 192 or 256, there's even less chance of detecting any differences with WAV.

    If you look at the frequency spectrum of MP3 files at bit rates from 192 kbps and above it's easily seen that the very high frequencies are stlll present, with only very limited curtailment. It's only when the bit rates are 160 kbps and below that the frequency curve begins to become truncated at the higher frequencies above about 16 khz. This is such a high level that most people wouldn't be able to hear them, even if they might like to think they ought to hear them, or be aware of them somehow.

    The fact is that one's sensitivity to high frequencies begins to decline as one ages. One doesn't need to be very old before anything above about 12-13 khz is lost, and this gets worse stiil for the 60+ down to about 10-11 khz.

    In any case, the way MP3 files chiefly lose size compared with the WAV original is not so much as by truncating the high frequencies but by masking (i.e. eliminating) sounds that one is unlikely to hear because some other sound is likely to predominat over it, in a process called "masking", based on a psycho-acoustic model of some kind.

    With Bigshot's test files, I wonder how the risk of inadvertent "cheating" can be eliminated if the test is carried out by the person himself/herself, rather than with the aid of an assistant. One only has to glance accidentally at the size of the file to see which has the higher bit rate. Another potential problem is that some media players show the bit rate and codec whilst the work is in play.
    Last edited by Partita; Jun-16-2019 at 12:30.

  12. #27
    Senior Member
    Join Date
    Sep 2012
    Posts
    1,118
    Post Thanks / Like

    Default

    Further to my post above, here's a useful test for your ears and your hi-fi.

    Take a piece of classical music you know well, and create a 2 minute WAV 16 bit file.

    Then apply a low-pass filter with a high roll-off, starting at 20 kbps. Repeat at 19 kbps, 18 kbps etc down to 10 kbps. Save each file as WAV.

    You'll have 12 files including the original unadulterated WAV file.

    Find some way of listening to each file without knowing which is which. Listen to each and rank them according to quality.

    You may be surprised at the results. Try it for different types of music.

  13. #28
    Senior Member Tero's Avatar
    Join Date
    Jun 2012
    Location
    USA
    Posts
    263
    Post Thanks / Like

    Default

    A lot of mp3 music (or iTunes mp4) sounds like crap because of loudness. There are no dynamics in music produced for the young folks of today.

    if I buy any Rolling stones CDs, the loudness is there in the re-releases. The first CD releases had some deficiencies, but the loudness was not there. Beatles CDs have enjoyed a bit less loudness treatment.
    https://www.youtube.com/watch?v=3Gmex_4hreQ

  14. #29
    Senior Member bigshot's Avatar
    Join Date
    Nov 2011
    Location
    Hollywood U.S.A.
    Posts
    5,866
    Post Thanks / Like
    Blog Entries
    1

    Default

    Quote Originally Posted by Partita View Post
    With Bigshot's test files, I wonder how the risk of inadvertent "cheating" can be eliminated if the test is carried out by the person himself/herself, rather than with the aid of an assistant. One only has to glance accidentally at the size of the file to see which has the higher bit rate. Another potential problem is that some media players show the bit rate and codec whilst the work is in play.
    The lossy samples are all contained within a single lossless FLAC file. There are ways to try and cheat, and people have tried, but it isn't possible to follow all the rules of the test and cheat without it being apparent in the results.
    CD Sound Is All You Need: http://xiph.org/~xiphmont/demo/neil-young.html
    AES Audio Myths Seminar: http://youtu.be/BYTlN6wjcvQ
    AES Damn Lies Seminar: http://youtu.be/Zvireu2SGZM

  15. #30
    Senior Member bigshot's Avatar
    Join Date
    Nov 2011
    Location
    Hollywood U.S.A.
    Posts
    5,866
    Post Thanks / Like
    Blog Entries
    1

    Default

    Quote Originally Posted by Partita View Post
    Then apply a low-pass filter with a high roll-off, starting at 20 kbps. Repeat at 19 kbps, 18 kbps etc down to 10 kbps. Save each file as WAV.
    I've done that test. It is pretty remarkable. Audiophiles spend tremendous effort optimizing the upper frequencies when that is the least important range in the sound spectrum.

    One thing to keep in mind is that the frequency range is logarithmic. The distance between 20 and 40Hz is one octave. The difference between 10kHz and 20kHz is also one octave. The end of each octave is approximately double the number of the start of it. So 10kHz to 20kHz is only a ninth of the audible range. It isn't half.

    Quote Originally Posted by Tero View Post
    A lot of mp3 music (or iTunes mp4) sounds like crap because of loudness. There are no dynamics in music produced for the young folks of today. if I buy any Rolling stones CDs, the loudness is there in the re-releases. The first CD releases had some deficiencies, but the loudness was not there. Beatles CDs have enjoyed a bit less loudness treatment.
    https://www.youtube.com/watch?v=3Gmex_4hreQ
    The Rolling Stones remasters are complete remixes in some cases. Their music was always recorded quick and dirty- basically live in the studio- and distortion was a big part of the sound. They also used slap back and wire reverbs and phone futz vocals that have a distinctive sound. When they remixed and remastered the albums for digital they replaced a lot of that with clean digital reverb and EQ that just doesn't sound right.

    If you want the way the Stones used to sound, look for the Mono Box. That uses the original mono mixes with minimal alteration in mastering.
    Last edited by bigshot; Jun-16-2019 at 19:04.
    CD Sound Is All You Need: http://xiph.org/~xiphmont/demo/neil-young.html
    AES Audio Myths Seminar: http://youtu.be/BYTlN6wjcvQ
    AES Damn Lies Seminar: http://youtu.be/Zvireu2SGZM

Page 2 of 3 FirstFirst 123 LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •