A question for the design geniuses hereabouts

[theloniouscoltrane]

Hello folks, I have a serious question. I have been using the Beta for a couple of months with very few problems. I am interested in having the best sound possible and I have just run across another player - the Amarra - that claims it outperforms all others because it can it can play back various bit depths and sampling rates. It sounds a little like voodoo, so I'm curious if those of you who know way more than I can explain this to me. Now, I don't have much experience with other players, but I've never noticed having a problem with Media Monkey using anything from cheesy Apple 128 m4a files all the way up to pristine Linn 24/192 files with bitrates at just under 6,000.

Amarra mostly compares itself to iTunes and they extol their ability to bypass several parts of iTunes, such as changing its playback path and also changing the sample rate of Apple's CoreAudio engine so that it's the same as the actual file being played. I wasn't aware that iTunes CoreAudio engine even existed, let alone that it converted the sample rate as it played, nor the implication that this conversion puts sufficient strain on a computer's operating system that it makes the sound suck.

If I take Amarra's word for it being true that this CoreAudio is constantly changing the signal so that it is converted to whatever makes iTunes happy, does Media Monkey do anything like that? (They imply that they have the only player that bypasses all of this muck, and of course, the engineering to perform this bypass is quite expensive at $695)

Even more, regarding the implication that only the Amarra bypasses any signal conversion and sends the signal straight to the digital sound conversion (RME FIreface 800 in my case). Is that true? Does MM down-convert a 24/192 flac signal and change it to a 16/44.1 wav file? Or something more nefarious? I've always used MM DirectSound output, assuming I was getting a direct sound output.

They also claim that since they perform DSD to PCM using "the advanced Phillips based Trellis conversion," their system sounds better. Their other superior methods include "automatic sample rate switching, double-precision audio processing, RAM-based Cache Mode for disk-free operation, Native FLAC playback & conversion and choice of dither algorithms."

Is this other player blowing some smoke at the well-heeled audiophile and hoping they won't notice? It sounds that way to me, but I don't have the engineering background to refute it. Or, if it is really doing something really great that MM doesn't do, can any of these improvements be implemented in MM?

[Otello]

First of all, Amarra is a Mac OS.X player, not for Windows.

Amarra solve a problem with iTunes, which cannot work in exclusive mode ("hog mode" for Mac users) and cannot automatically switch the sample rate according to the file. Period.
If you manually change the sample rate in MIDI/Audio setting according to the file you are playing, you get exactly the same result, i.e. bit-perfect output.

Media Monkey 4 may work in exclusive mode.
With Windows 7 (or Vista) select WASAPI output plugin, then click on "Configure", check "exclusive mode" and select "Automatic choise of format (per track)".
This way MM4 has a bit-perfect output and automatically switch the sample rate according to the file you are playing.

[theloniouscoltrane]

Thank you so much. I completely missed the Mac-only part. Silly me.

But also thanks for reading through to the real intent of the question. I'll try switching over today.

[Pietercape]

Hi,
Interesting post and response. I listen to my classical music through a Cantatis Overture 192 soundcard and OS is Vista 64 bit with MM4 and MM Direct Sound selected.. The soundcard came with software that supresses Windows sound management as I understand it. After reading this post I switched to the WASAPI output plugin as suggested in the post.

I normally use the Windows Volume Mixer to set the soundcard playback volume at around 85% as recommended by the soundcard manufacturer and the MM output volume at around 15% for earphone listening (Sennheiser HD480 Classic II) and at around 50% when I use my Amp/Speaker setup.

With WASAPI I lost all control over volume be it the soundcard output or using the Windows Volume Mixer. Even the MM volume control on the minimised player bar does not work.

It "seems" that the WASAPI sound is crisper/cleaner but the sound is at very high volume and so I cannot be sure.

Any suggestions on how to control the volume other than Levelling Track Volumes?

[Otello]

Hi,

I never tested your sound card, but according to the manufacturer FAQ, at the very end of http://www.cantatis.co.uk/index.php?pag ... tions.html, WASAPI is is NOT supported:

Note that as the Overture 192 audio driver is a highly reliable and well tested WDM driver it does not support the new WASAPI interface that allows for "exclusive mode" access to the device driver so you must select the desired sample rate and bit depth using the Overture 192 Control Application and Windows playback devices control panel as described above.

[Otello]

P.S.

By the way, WASAPI output is meant for serious audiophiles, people looking for the best possible sound from a PC connected to a hi-end audio system.
The goal is a "bit-perfect" output; i.e.: the DAC receives the digital data without any modification by the program or the windows volume mixer.

Now, using WASAPI may not be enough; to get a bit-perfect output, you cannot:
- convert the sample rate
- use equalizer, DSP, normalizer, reply-gain, etc.
- use the digital volume control.

[Pietercape]

Hi Otello,

Thanks for your response. It is a pity that the Cantatis does not support Wasapi, it would have been interesting to compare the sound at lower volume!

I have had my Cantatis for two years now and I am very happy with it. It was pricey but I guess you get what you pay for?

[jametuta]

Hi,

That affirmation from Cantatis doesn't make any sense at all, by definition any WDM driver should work in Wasapi mode.

Being an owner of one Overture 192, it always worked in WASAPI or ASIO (with asio4all) in my media player software.

Regards
PT

[Pietercape]

Hi Jametuta,
Yes the Cantatis Overture does play the output from Wasapi but in my setup I cannot control the volume. My card is set to use a bit-rate of 24 and the sample-rate is set at at 96000k.

I asked Cantatis what happens when I use Wasapi and the Overture 192 under these conditions and the response was:

1) 16-bit source played at 24-bit. What happens with the Overture drivers and soundcard is that the bit-depth is maintained. This means that the sound is output at 24-bit, but actually the lower 8 bits are ignored.Some cheap systems will try to "guess" the extra bits (and we could do this too), but this is not sensible in our view and would/could majorly change/destroy the sound - which is not the aim of the purity of the Overture, so the 16-bit is maintained, just output in 24-bit format.

2) 44.1k source played at 96k. This is upsampled. The original sound will gain extra information in the upsampling, and the effect will to be give the impressions of greater depth, high highs and to "sparkle" more.With a 44.1 sample rate the max frequency that can be played is 22.05kHx (half the sample freq). at 96k the max freq is 48kHz. Now, this is in the inaudible part of the frequency, but there are effects that this does have back in the audible range. This is always a topic of personal taste and the upsample algorhythm we use is one of the most complex in the industry. Some people do crude upsampling and the sound comes across as over bright and artificial. We've used some of the latest upsampling techniques to ensure that the sound is not massively changed.

They say that they are not supporting Wasapi at this time but that they may do so when they are satisfied that the issues that MS has with Wasapi are resolved. They are MS Development Partners and I suppose that they will know when to change their position.

I will wait patiently!

[theloniouscoltrane]

2) 44.1k source played at 96k. This is upsampled. The original sound will gain extra information in the upsampling, and the effect will to be give the impressions of greater depth, high highs and to "sparkle" more.

This seems physically impossible. Years ago in the hifi world, they used to hoist this old concept. If you take a painting and put it in a frame that is twice the size of its current frame, you wouldn't get a picture that was twice as big. It would just have a very large frame around it.

[Pietercape]

Hi,

The quote is from sound experts and Microsoft Development Partners and they are in the business of building sound cards and specialised sound servers for purveyors of digital music.

Who am I to argue?

[rovingcowboy]

Amarra mostly compares itself to iTunes and they extol their ability to bypass several parts of iTunes,

Your line there tells me to stay away from it. i don't care what they can do. when apple changes its code they will be
out of luck and need to try another work around. this to me is just a red flag to all the apple guys to change their code
and they will you know that.

[Otello]

2) 44.1k source played at 96k. This is upsampled. The original sound will gain extra information in the upsampling, and the effect will to be give the impressions of greater depth, high highs and to "sparkle" more.With a 44.1 sample rate the max frequency that can be played is 22.05kHx (half the sample freq). at 96k the max freq is 48kHz. Now, this is in the inaudible part of the frequency, but there are effects that this does have back in the audible range.

I'm sorry, but this is wrong.
Upsampling do NOT add extra information, it just interpolate samples; also, there is no music recorded over 20 KHz (given that microphones may catch these frequencies and your speakers may reproduce them).
Upsampling is used because it allows to filter at a higher frequency (Re: http://en.wikipedia.org/wiki/Nyquist%E2 ... ng_theorem), so the filter may be less steep, so you get less phase rotations in high frequencies, so you get a better sound in high frequencies.

By the way, it is much better to upsample by and integer (i.e: 44.1K x 2 = 88.2K; x4 = 176.4K)