2014-10-07

Audiophile music - 10 conclusions

I am passionate about sound quality lately. Here are my early conclusions from the research and experiments I did. This is the order of sound optimizations that you could try in order to improve the music experience.

1. Live music sounds better than recorded music

2. Un-amplified live music sounds better than amplified music

3. Recorded music sounds better on loud speakers than on headphones

4. Expensive headphones (>100$) sounds better than cheap ones

5. Open-air headphones sounds better than closed enclosure ones

6. Loss-less music (like FLAC) sounds better than MP3

7. High bitrate MP3 (>128kbps) sounds better than low bitrate

8. A professional DAC (or sound card) sounds better than a cheap one

9. High sample rate music (>44KHz) could sound better than 44KHz (or not)

10. Music encoded at 24bits could sound better than 16bits music (or not)

 I will say a few words about each, from my limited experience and research:



1. Live music sounds better than recorded music
If you want a full music experience, try live. Even amateurs musicians sounds betters for me than recorded music. The live sound is more natural, the harmonics are all there. There is also the psychological experience: you cannot know for sure what will happen to the next note. Recorded music will always sound decent and without mistakes. Live music is always a bet that all will go well, that no mistakes are made. Just look at the artists in an orchestra, they are very tense to play exactly the right note at the right time. The experience is more full when you see the artist in front of you, especially if all goes well enough ;) Some great singers might even adapt the performance to the  mood of the audience, making it an irreproducible experience. Unfortunately such live experience is rarely feasible to achieve.
2. Un-amplified live music sounds better than amplified music 
Amplified music tends to remove some of the naturalness of music, but for some reason it tends to still sound better than recorded music. It's the same experience as above regarding the miracle that music continues. However, there is a small delay of music in the amplifiers, sometimes the amplified music looks like playback. Even worse, you could hear the un-amplified and amplified music in the same time at different delays. However, this music still sounds a bit better than recorded music. It's not only the psychological experience, the equipments in a concert hall are usually better than the equipment used for playing recorded music. The music quality tends to stay higher if the audio chain is short. It is also possible that, despite all the improvements and proves, a single analog amplifier to be more precise to reproduce sound than an analog-to-digital-to-analog chain. On the other hand, you cannot be too often in the same room or concert hall with your favorites singers.
3. Recorded music sounds better on loud speakers than on headphones
The headphones tends to create the sound "in the middle of the head". There are some techniques to make the sound more spacial. However they are usually only tricks that don't give the full spacial experience of decent speakers having some space in between and between them and you. There are a lot of room reflections that creates the experience of space. Also, "virtual 3D" effects could damage the music quality.

It also helps to have an audio system composed of multiple speakers specialized on different frequency ranges. Depending on the size of the speaker, it can better reproduce some specific frequencies (big speakers for bass, small speakers for treble). The membrane material may be different too. Just imagine that you have a perfect electrical audio signal and it tries to move the membrane of the speaker accordingly. The membrane material will have different elastic response for different frequencies, and many reflections of the initial move. It is almost impossible to have a perfect speaker, but some are better than other.

From the spatial point of view, my car audio system sounds better than good headphones. However, good headphones sounds better than mediocre speakers, at a lower price.
4. Expensive headphones (>100$) sounds better than cheap ones
Usually the best investment you can make to improve the quality of the sound is a pair of decent headphones. The most damage on the sound is usually done when transforming the electrical signal to actual sound. The headphones could have bad interferences on specific frequencies or can reproduce the level of different frequencies very unbalanced. It is very hard to make perfect headphones. "Decent" starts from around 100$, however more expensive does not mean better. Check the reviews on the "audiophile" forums, but try to have more sources in order to avoid masked advertising. Usually you should go with companies that have a history on the professional sound industry and not for headphones advertised by celebrities. There are many companies with good reviews on headphones. I only tested (and liked) Grado, Sennheiser, Audio-Technica and Creative. Of course, each company have entry-level products and more professional product. From my research, after a price (let's say 300$?) you won't notice much difference. Even worse, higher fidelity headphones could overemphasize the small imperfections on the sound source.
5. Open-air headphones sounds better than closed enclosure ones
Most of the headphones are closed on the outside, in order to isolate from the outside noise and to not leak too much on the outside (like in an office). However, this makes the sound to reflect a lot and produces a sound that seems to came from a barrel. If you will try open air headphones you will notice a more "open" sound. The bass is a bit less powerful, however you will notice a lot more details in the sound, like it's "more stereo". This is because the sound will be less distorted by reflections. I highly recommend open air headphones, but not for office. The sad thing is that they tend to be significantly more expensive, even when the only difference is a hole in the plastic.

6. Loss-less music (like FLAC) sounds better than MP3
The music source is of course very important for a good music quality. A good source of digital music is a CD digitized from a professional studio recording. However, one full CD can hold around 60minutes of music, that is very little.
MP3 is a very popular format for storing digital music, as it takes like 10 times less space than the uncompressed CD format (PCM - think WAV). However, the MP3 music compression is "lossy" in the sense that the music will not be identical with the CD source. This is unnoticed most of the time. Most people will not hear a difference starting with 128-256kbps MP3.

However, on a high fidelity audio system you want to have the best input quality you can get. For this you can use a "lossless" compression like FLAC that stores the music in less space than the original CD tracks but it gets uncompressed to the exact bits from the CD. Using FLAC the music will be exactly as it is on the source CD and you can even check this (see accurate rip).

Unfortunately you can find many FLAC files that are created from a poor source (like MP3). In such cases the FLAC will preserve the imperfections of the MP3 source and will only take space without any quality gain. Try to find FLAC files created from very good sources (like original CD). You will also find FLAC files created by digitizing vinyl records. Many people appreciate the sound of vinyl over CD music, probably because it have a specific color of the sound. The digitized FLAC could presumably preserve that color, however it will contain also the vinyl imperfections. At least it will not be affected by more scratchings...

7. High bitrate MP3 (>128kbps) sounds better than low bitrate
 There is a noticeable quality difference until 128kbps MP3. Between 128kbps and 256kbps the differences are less obvious. You can go up to 320kbps where most people will not be able to differentiate from the original source. However some are still able to hear the difference on some passages. Your mileage may vary... ;)

Over 128kbps you should have a very good sound source, otherwise it will not help. Very bad sources (like old tapes digitized) could sound a bit better when encoded on lower bitrates, because encoding tends to extract sinusoids from the source and ignore details that are not sinusoids (noise). Higher bitrate means also a better representation of the original noise from a damaged recording.

8. A professional DAC (or sound card) sounds better than a cheap one
The digital sound is always transferred and uncompressed in the same way on any system. The difference starts with the conversion from digital to analog, from bits to electrical signal. This is done by the sound card usually but more generic such system is named DAC (Digital to Analog Convertor). The quality of the DAC could put a heavy footprint to the sound quality. Good DACs must be very faithful to the digital input. Also, the analogical audio path should have very low distortion, low noise and good frequency linearity.

In practice, good DACs will be advertized as supporting high sampling frequencies and bit depth (like 96-192kHz and 24bits), however, in fact, their quality (where present) comes from the quality of the audio path, low noise/distortion and good linearity. 

Also, it is possible that a regular sound card will not have enough power to drive the headphones. Even it it sound loud enough, at some frequency there might be not enough energy to reproduce the sound, resulting in lower quality. It is a known rule that a powerful amplifier at low volume sounds better than a medium power amplifier at the same volume level. Normally you should have a headphones amplified after the DAC (sound card). However, there is a good compromise to get an USB DAC that has a headphone amplified included. The only disadvantage is that the signal is not very good to input a speaker amplifier.

There is a big debate on this and often exaggerations on the benefit of different expensive audio systems. At this moment I am skeptical about improvement by very expensive cables for example. You could have a signal issue if you have a bad electrical contact or really bad cables, but I don't think that copper purity would affect the audio experience. For the electrical contact problem it might be better to have high impedance impedance headphones - however they require more power.

Amplifiers with tubes sounds like a good idea if you can afford the money. I have a limited experience with it, however my impression is that they could provide a more colorful sound, lower noise and better signal clipping. However, usually you can use the money more efficient to improve the other parts of the audio chain, like buying better speakers/headphones.

Despite the marketing tricks, I noticed, however, that a good quality DAC sounds a bit better even with MP3 at 44.1KHz/16bits.

9. High sample rate music (>44KHz) could sound better than 44KHz (or not)
Audio professional says that 44KHz sampling rate is enough to extract from the analog signal all the human audio spectrum (16-22KHz). Bigger sample rates (like 96KHz or 192KHz) are useful to digitize ultrasonic sounds, but this is only useful if you are a bat :) Also, 16bits should be enough to represent the smallest changes in amplitude that a human is able to detect.

However, when digitizing the music, there are technical reasons for using a higher sample rate (like 192KHz). After the signal is ready, it should be safe to convert that music to 44.1KHz/16bits (CD quality). Usually the reason that higher sample rate sounds better than lower sample rate is that this music is for audiophiles (more expensive) and they invest more care when preparing the master in the studio.

There might be a reason to use 48Khz in order to improve the stereo perception of the music, but this is the highest sampling frequency that is technically required and proved by ABX tests.

Let's keep a reserve here, maybe higher sample rates could have some use in representing very subtle nuances that are not sinusoidal in nature (a note's attack). But you will not be able to notice it! I found several 44.1Khz/16bits audio files that sound better than most 96/192Khz 24bits files I have, so better sampling rate is not a guarantee for better quality. You can actually make the sound worse if the audio path does not know to handle the higher frequencies, it could introduce distortions or aliasing. I would say that it's enough to use really good quality audio FLAC at 44.1KHz/16bits.

Music encoded at 24bits could sound better than 16bits music (or not)
Human ear should not be able to detect sound differences at more than 16bits depth. When processing the music, however, it is useful to use 24bits for preventing audio clipping when mixing. If the audio gain is correctly adjusted you could safely downsize the music to 16bits without noticing a difference.

The music available on 24bits could sound better because it was more carefully digitized and mixed, but this is only a correlation. I personally found that some audio cards sound even worse in 24bits than on 16bits because of software or hardware bugs. It is safe to keep a reserve, but most probably you won't need more than 16bits to have the best (digital) audio quality you could get.



The 11's element

There is another factor along this 10. The music you like will give you a better experience no matter what. Especially if you have a good memory of that music, it's enough to hear something even remotely similar to recreate that great music experience from your memory. It's not the same, however, for music that you don't know and you would have liked. A bad audio system could prevent you to really enjoy that new music.

Don't invest too much on the audio system. Just make your own experiments and decide what's good enough for you. I hope these ideas helps a bit.

2 comments:

  1. The perceived sound versus recorded sound is such a complex topic ! When you hear a sound, and you instinctually turn your head toward that sound…. how do you know where it comes from?
    Our ability to perceive sound direction works through a process known as binaural hearing, which essentially means “hearing with two ears”.
    As sound travels though the air at a high speed, it still reaches each of our ears at two different times and the brain is able to translate this small delta - this time difference in spatial location. This is why live music cannot be compared with recorded music, as each instrument and voice has its own position related to the position of the listener.

    Please listen to this "binaural sound" on youtube in HD using a "good" pair of headsets - is important to use headsets as the sound reaching each ear must be isolated from the other : https://www.youtube.com/watch?v=u163wC6mP2A&list=RDitLxXeyM2aM

    Saadon

    ReplyDelete
  2. The "binaural sound" is the argument to use a sampling frequency higher than 44KHz, like 48KHz or 96KHz. However, when using more than 2 microphones, it is unlikely that mixing the sound could recompose the stage for a perfect binaural experience.

    ReplyDelete