[Douglas256]

fewtch,

I'm glad you were able to hear a difference between 16bit/44.1kHz and 24bit/96kHz on your system. Clearly, I'm a big fan of recording at 24/96 (or 24/48). That said, a good conversion to 16bit/44.1kHz that is then upsampled/oversampled during play back can close the gap quite a bit between 16/44.1 and 24/96. While I still hear a difference (foobar2000 -> S/PDIF -> Benchmark DAC1 -> HeadRoom Max -> HD 600), it isn't nearly as dramatic a difference as when you record at only 16/44.1.

Since most professional recordings are mixed at 24bit and 48kHz, 88.2kHz or 96kHz, I'm surprised more recordings aren't being released on DVD-A. While I haven't actually looked into the reason, I would guess that the very high cost for the software to encode DVD-A and the higher price for the media is the cause.

 

[fewtch]

Quote:

Originally posted by Douglas256 fewtch, I'm glad you were able to hear a difference between 16bit/44.1kHz and 24bit/96kHz on your system. Clearly, I'm a big fan of recording at 24/96 (or 24/48). That said, a good conversion to 16bit/44.1kHz that is then upsampled/oversampled during play back can close the gap quite a bit between 16/44.1 and 24/96.

Oh, I definitely agree. In fact, a good upsampling from original 16/44.1 source material to 24/96 is a great sonic improvement to my ears. I've only done it with my soundcard so I'm not sure how much of the improvement comes from simply outputting in the DAC's native format using really good upsampling software (rather than letting the card handle any conversions)... but the improvement is definitely there to my ears. Some of my CDs (err, lossless encoded files) sound downright "analog" at times, and there's a sense of hearing much deeper into the noise floor. I think the SQ would come as a real surprise to many around here who didn't know they were hearing a $150 soundcard

.

 

[tigre]

Douglas256, thanks for the samples.

Fewtch: As you probably know, I'm interested in finding out if there's a audible difference between 24/96 and (state-of-the-art) 16/44.1.

Quote:

Both played back through Senn HD-600s and a Headsave Ultra hooked to an M-Audio Audiophile 24/96 sound card.

Could you please tell what player you used for this comparison?

Have you tried to upsample the 16/44.1 to 24/96 again and compare to the original?

Or even better: Could you please try Shibatch Sampling Rate Converter (SSRC) (from Rarewares) with KikeG's .bat file (details here)?

(It automatically resamples 24/96 to 44.1/16 using fast mode (=less pre-ringing) + soft ath noise shaped dither and back to 24/96 using flat dither + slow mode.

Usage: given the .wav file's name is test.wav + all files (SSRC, .bat, .wav) are in the same folder, type
ssrc_down_up_dath test

This way you'll ensure that the difference is not caused by the player and/or the soundcard's performance varying on different sampling rates / bit depths.

 

[Music Fanatic]

Quote:

Originally posted by theaudiohobby I see that you coyly avoided answering my question as to why on DVDA, the software vendors are not playing ball. Oh well..., I suppose I already answered it myself, after taking into account all the issues, it did not make financial sense for the software vendors to invest in it.

Why do you think it would be different for the software vendors of SACD recordings and the software vendors of DVD-Audio recordings (true, now SACD holds quite a strong lead, but originally, DVD-Audio had the lead.)

I'm guessing (although you did not explicity say this) that you would think that both groups were rational business decision makers. How did they reach such different answers?

 

[theaudiohobby]

Quote:

Originally posted by Music Fanatic I'm guessing (although you did not explicity say this) that you would think that both groups were rational business decision makers. How did they reach such different answers?

There is no need to let this drag on, Suffice to say that no matter how irrational your original business decision, a continuous flow of red ink will make you sit up straight and think rationally.

 

[tigre]

Since my previous post I've read the last few sites of this thread. Some people here seem to be interested in finding out if 24/96 sounds better or not (and how much), so here you go:

Here's the 24/96 vs. CD quality test thread, here some 24/96 samples (and an easy + high quality way to convert 24/96 -> 16/44.1).

Basically the test works like this:

1. Download original (wavpack losslessly compressed 24/96) and decode to .wav, decoder avail. at rarewares.

2. Use SSRC + bat file to convert to 16/44.1 and back to 24/96. (Details see my previous post).

3. Compare. (For those who believe in ABX

: WinABX and ABC/HR have problems wiht 24/96 playback on Win2K (probably XP too), foobar2000's ABX tool works perfectly).

(4. If there are any differences it'd be interesting if
only 16/44.1 file sounds different (-> difference caused by player/soundcard) or
both resampled files sound different compared to the original (-> the difference is most likely really caused by bit depth/sampling rate). )

I'd be glad if everyone who tries would post the results.

BTW: Douglas256, would it be OK for you if I uploaded your samples to the Hydrogenaudio thread?

 

[Glassman]

thanks for the samples and also thanks for the link to HA.org..

I performed some short tests and to be honest I can't tell much difference

each original file resampled down to 16/44 dithered by foobar and then played together with the original using 192kHz resampling and triangular dithering..

maybe my source sucks, maybe my phones sucks, maybe my ears sucks or maybe there's really not a lot of difference..

the only place where I heard some difference was the beginning of the Mangione07.wv

 

[fewtch]

Quote:

Originally posted by tigre Douglas256, thanks for the samples. Fewtch: As you probably know, I'm interested in finding out if there's a audible difference between 24/96 and (state-of-the-art) 16/44.1. Could you please tell what player you used for this comparison?

WinAMP v2.91 (on Win98SE) with Peter's WaveOut v2.0.2a SSRC plugin, using "slow mode" for the resampling, triangular dithering, triangular distribution. Output on the Audiophile soundcard was "directly" to the WavOut 1/2 analog port, nothing in between like the Monitor Mixer or anything else (no DSP, no WinAMP equalizer, no 'Microsoft Sound Mapper', nothing).

I just gave a quick subjective impression (this not being HA, and me being lazy

) so take it as you will...
Quote:

Have you tried to upsample the 16/44.1 to 24/96 again and compare to the original?

No. But these days I listen to my regular 16/44.1 files upsampled in realtime to 24/96, and do think I hear a difference. To ABX this would require quite a bit of time to get a sampling across a wide variety of music, and I would want to use longer samples over a longer period of time... more like an "ABX project" than a few ABX tests. Just too lazy...

 

[tigre]

Quote:

Originally posted by Glassman thanks for the samples and also thanks for the link to HA.org..

You're welcome.
Quote:

the only place where I heard some difference was the beginning of the Mangione07.wv

I'm desperately trying to hear a difference, so far without success. Could you please describe what the difference was like - so I could have a chance to hear it too?

 

[tigre]

Quote:

Originally posted by fewtch WinAMP v2.91 (on Win98SE) with Peter's WaveOut v2.0.2a SSRC plugin, using "slow mode" for the resampling, triangular dithering, triangular distribution. Output on the Audiophile soundcard was "directly" to the WavOut 1/2 analog port, nothing in between like the Monitor Mixer or anything else (no DSP, no WinAMP equalizer, no 'Microsoft Sound Mapper', nothing).

I haven't used WinAmp for a while, so I might be completely wrong on this ...
IIRC Winamp truncates to 16 bit on every processing step (e.g. DSP) - and I'm not sure if it is capable of 24bit output at all. So this might be the reason for the differences you've heard.

I've tried to ABX the glassofwater sample using the method described before (fb2k's ABX tool used). I though I heard subtle differences in high frequencies and in the noise at the beginning of the sample, but ABXing wasn't even near success, so probably it was imagination.

(Aureon 5.1 Sky (24/96) -> Sennheiser HD 540 used)

 

[Glassman]

Quote:

Originally posted by tigre I'm desperately trying to hear a difference, so far without success. Could you please describe what the difference was like - so I could have a chance to hear it too?

right now I've tried to ABX this but without any success.. it was most probably my imagination..

I'm going to concentrate on the glass of water now, but any suggestions for audible parts of the files are wellcome

FOR GOD'S SAKE, I MUST HEAR A DIFFERENCE!

edit: now I've failed with PouringAGlassOfWater too.. seems like CD is definitely not the limit if mastered and played right..

 

[Douglas256]

tigre,

Feel free to post/upload my samples over at HA, just be sure to cite me. If you want any more samples (and have suggestions of what you want me to record), I'd be more than happy to help out.

As for determining if there is really a difference between the origional 24/96 and downsampling to 16/44.1 and then upsampling to 24/96 again, I can't point my finger at any part of the recording and say, ``Ahh, it sounds different there!'' When I ABX in foobar2000 with my Ety's, I always end up around a 30% chance of guessing. Sometimes I think I hear a difference at the end of the pouring a glass of water, sometimes I don't.

The important thing, in my opinion, is recording at the higher bit depth and sample rate. I hear a noticable difference between recording at 16/44.1 and 24/96 (which I usually process as a 24/96 and then downsample to 16/44.1 before saving -- just as I scan my photographs at 16-bit per channel, color correct, and then save at 8-bit per channel).

Of course, this is all a bit of a mote point given how much compression (in the bring the quite parts up and the load parts down sense) is used on all the CDs we buy today. Except for my Ani DiFranco and "audiophile" CDs, most of my CDs have 80% of their music in the top 10 dBFS, 90% in the top 15 dBFS and 98% in the top 20 dBFS. So, there isn't much reason to have a medium with a dynamic range of 120+dBFS when all the currently mastered disks only have information in the top 25-30 dBFS. I guess it's time I start calling up record companines and complaining about how crappy their disks sound again.

 

[tigre]

Thanks for the sample posted over at Hyodrogenaudio. The last water sounds at the end of the file sound a bit artificial to me, most likely because of noise removal. Under 'real life' circumstances it'd be a good idea to remove less noise to avoid this kind of artifacts IMO, but for demonstrating a very low noise level (and to get an impression of how loud 16 bit dither noise sounds compared to ~ full scale signals) this sample is good.

 

[Douglas256]

tigre,

Now that I carefully listen to the end of the file I posted over at HA, it does sound a bit artificial (at least for the last couple of little water bubbles popping). I guess I need to get my act together and treat some of my rooms : ).

Since I posted the new sound over on HA, I might as well put the same file here.

Pouring a glass of water with the background noise removed using Cool Edit Pro

This time, I used Monkey Audio to compress it. (I'll probably go back and do the same with the previous files.)

 

[Glassman]

read the second paragraph here called Demo2: different sampling rates

http://www.hydrogenaudio.org/index.p...pic=9311&st=51

this explains everything.. and from what I understood they were swithing the ADC's samplerate, not the DAC's, so recording 24/192, editing and then putting as 16/44 on CD that will be resampled back to 24/192 at playback will do just fine - no DVD-A or SACD needed!