Test Files 24bit Vs 16bit, And 96khz Vs 48khz, How different do they sound? Options

[MLXXX]

There don't seem to be too many test samples on the net of this kind. I have prepared two test samples as follows:

Test file 1: 96KHz/24bits followed by 96KHz/16bits; or it could be 96/24 then 96/16
Test file 2: 96KHz/24bits followed by 48KHz/24bits; or it could be 48/24 then 96/24.

If you have built your HTPC and loaded your player software and are not sure whether you are getting 24-bit sound rather than 16-bit sound, or 48KHz sampling rather than 96KHz sampling, these short test files should give an idea of the differences to expect. This is a topical subject for people who are trying to get Cyberlink Power DVD not to limit the bitrate to 48KHz and the bit-depth to 16-bits when playing a Blu-ray or HD-DVD disc.

The first test file compares bit-depth. It's the sound of a dying chord produced by an upright piano. It is repeated after 6 seconds. Your task is to listen and report whether the repeated version and the original version sound the same, and if so, whether one sounds better than the other. Also how great the difference is to your ears between the two versions. (The poll questions ask for these details.) Here is the first test file:

FadingPiano-96KHzWave-- one at 24 bit and other at 16 bit dithered.wav

Please use 'save target as' rather than immediately playing the file, as you may wish to listen to it again at a later time, and download bandwith is limited. The file is 6.23MB.

The second test file is the same piano chord but the difference between the original version and the repeat is in the sampling rate. One version is 96KHz at 24 bits, and the other is 48KHz at 24 bits. Here is the second test file:

FadingPiano-24bitsWave-- one at 96KHz and other resampled to 48KHz.wav

Of course, the player that you use to play the wave file and the sound card on your pc, need to be capable of operating at 96KHz and 24-bits. So there is a certain amount of "chicken and egg" about this exercise! The reason you hear no difference may be because your setup is for some reason operating at only 16-bits.

I'd also suggest you choose a time when you are rested and relaxed, and mentally alert.

If the versions sounded different but neither version sounded 'better' than the other, than only complete the part of the poll that asks how different the versions sounded.


Methodology
Upright piano, recorded with a matched pair of Rode NT1-A microphones, at 96KHz and 24-bits.

A short extract (5.3 sec) of the wave file was then subjected to processing, using N-track Studio (version 4.2.1) as follows:

1. For the first test file, resampling to 16 bits using simple 1-bit dither. (This was done in preference to truncation, as to my ears truncation gives a dry sound whereas the dithered version sounds more like the 24-bit original.)
2. For the second test file, the original extract was resampled to 48Khz 24bits and then resampled back to 96KHz at 24 bits. No doubt some algorithms exist that might have impacted a little less on the sound quality than this method.

After a reasonable period of time, I will post (using concealed text) which way around each test file was assembled. It will be interesting if it turns out that people prefer the processed versions, i.e. preferring 48KHz sound to 96KHz sound, or preferring 16 bits dithered to 24bits. (IMG:style_emoticons/default/blink.gif)

This post has been edited by MLXXX: Feb 24 2008, 12:38 PM

[java]

First file end sound more interetsing and full. Second file prefer beginning.

Not much differences.

[MLXXX]

I've found some test files that may be of interest, at the bottom of this PC ABX Web Site webpage.

The castanets are definitely crisper in the 24/96 reference version, than the version downsampled to 44.1KHz and presented at 24/96.

I also found that using n-Track Studio and reducing the bit depth on the 24/96 reference sample to dithered 16 bits, made a noticeable difference even when maintaining the bitrate at 96KHz.

_________________

For those who like to use the ffdshow audio processor filter, it will resample on the fly, and can drop 96/24 down to 48KHz but may force a bitdepth reduction to 16 bits. There is a momentary disturbance when changing the resampling setting.

This post has been edited by MLXXX: Feb 25 2008, 01:38 AM

[MLXXX]

~ After a reasonable period of time, I will post (using concealed text) which way around each test file was assembled. It will be interesting if it turns out that people prefer the processed versions, i.e. preferring 48KHz sound to 96KHz sound, or preferring 16 bits dithered to 24bits. (IMG:style_emoticons/default/blink.gif)

Only 4 people replied.

1 person thought the original and repeat sounded the same, for both test files.
2 people thought there were slight differences within each test file and expressed preferences.
1 person thought there were clear diiferences within each file but expressed no preferences.

The preferences expressed were:
1st test sample ( bit depth) - both votes were for the repeat as sounding better
2nd test sample (bit rate) - one vote for the original, and one for the repeat.

And what was the order of the files?

Well the order in the first test file was: 16-bit (with basic dither) then 24-bit. [select with mouse to reveal]
And for the second test file it was: 48KHz sampling then 96KHz sampling.

There is no correct answer as to how great the differences were. Some people would suggest the answer is there were no significant differences unless they can be detected in double blind testing. In that connection I can highly recommend 'ABX' testing software, such as that found in a complete installation of the freeware foobar2000 (audio player with an ABX option). It is amazing how we can think we hear a difference but when actually put to the test, we cannot reliably do so. The mind plays tricks.

Many people claim that the difference between a 24-bit original and a 16-bit derived version cannot be detected if proper dithering is carried out to create the 16-bit version with music at realistic listening levels. I have been doing some tests of my own but am not hopeful, at this stage, now that I have discovered dither methods that add very little detectable noise.

As this topic is perhaps a bit specialised for our Digital Television Forum, I have made some posts in an audio forum, Hydrogen Audio, under Listening Tests, 16 bit vs 24 bit, any samples that work? beginning at post #41.

[drsmith]

1 person thought the original and repeat sounded the same, for both test files.

Having just listened to the test samples I too find myself in this camp.

[MLXXX]

Dr Smith, at least in relation to 24 bits vs 16 bits, your impression appears to be the same as that of audiophiles who have dedicated themselves to this kind of listening task. It appears that no-one has been able to come up with a 24-bit sample of music that sounds different when converted to 16 bits with dither, unless that music is at a very low recorded level, or is played back at an uncomfortably high gain and sound level.

Many people claim that the difference between a 24-bit original and a 16-bit derived version cannot be detected if proper dithering is carried out to create the 16-bit version with music at realistic listening levels. I have been doing some tests of my own but am not hopeful, at this stage, now that I have discovered dither methods that add very little detectable noise.

Well the collective wisdom of the HydrogenAudio forum is that a properly dithered 16-bit version of a 24-bit original cannot be distinguished, other than at extremely high listening levels.

The advantage of 24-bits is marginal, as explained on 26 March 2008 by "2Bdecided" in a "hydrogenaudio" post:

... If you really want to maintain the transient peaks of the waveforms (if only the music industry did!), and you want a huge amount of subsonic bass (ignoring the dedicated channel available for that for one moment), and you want dialogue at a reasonable level, and you want the noise floor below that of a dedicated listening room at all frequencies, and you don't want to have too much noise shaping in there because there are several stages of digital processing (lossy coding, level matching, delay, speaker EQ, room EQ etc), and you don't really trust all the equipment to be bit perfect, and you want the option of applying DRC to the output or not as you choose, and you want to match source levels in the digital domain without compromising headroom, then you probably want to start with more than 16-bits, and keep more than 16-bits throughout. (You also need a pretty amazing amp and speakers, not to mention very distant neighbours!)

So 16-bits are enough, but it's conceivable that you could throw together a situation where they're not.

Given the cost, which with today's disc media and hardware is so small as to be irrelevant, there's no reason to use "only" 16-bits on new disc media, even though in most situations 16-bits is more than enough.

This post has been edited by MLXXX: Mar 29 2008, 02:29 PM