[grokit]

I think one possible reason that those WAV/AIFF vs. FLAC/ALAC arguments never get resolved is because of inconsistent results between different computers. Everyone's mileage will vary according to how much processing power is available to uncompress those FLAC/ALAC files in real time, and how much processing power is actually available to do so. If the processing is bottlenecked, then errors are possible in the resultant signal timing stream.
 
If you have a kickass computer with tons of RAM, you probably won't notice any difference as there will be no bottleneck, but if you are trying to do it on a outdated cheap laptop (for example) it's quite possible that the overtaxed computer could introduce timing errors, particularly if your DAC is is relying on the computer to be the master audio clock because it is using adaptive USB audio protocols. Conversely, if your cheap old laptop doesn't have to uncompress the WAV/AIFF file then there would be less of a bottleneck when processing it and the computer will be less prone to timing errors. At least that's the way I understand it.
 
As far as jitter goes, geniuses like Gordon Rankin (who developed the driver-less asynchronous Streamlength receiver code), understand that if you move the master audio clock to an outbound device you can eliminate timing errors generated by that overtaxed computer, and he did it within existing USB audio protocols. This results in extremely low jitter and eliminates the need for resampling and reclocking. 
 
Streamlength is a unique and elegant solution that is being licensed to more and more higher-end devices these days and his own line of DACs have a great reputation for sound quality. So I will take his word for it when he says, "software changes the character of sound". He does have a product to sell, but he has nothing to gain by saying that. He explains his clocking technology much better than I ever could, since we're posting links:
 
http://www.usbdacs.com/Concept/Concept.html

 

[Dublo7]

So does anyone have suggested settings for Fidelia (including advanced options)? At the moment, I'm finding it to be quite harsh.

 
 

 

[JIGF]

 
Quote:

So does anyone have suggested settings for Fidelia (including advanced options)? At the moment, I'm finding it to be quite harsh.
 

It sounds harsh to me as well (with headphones), turning off all the "extra" settings (dither and everything to do with iZotope mainly) ameliorated it a bit for me.

 

[iamoneagain]

For anyone using Audirvana or Decibel and wants to be able to use Apple's iPhone Remote app, there is a solution.  It works pretty well.  The remote app loads the whole album into the player's playlist and then you control it with the regular handheld Apple remote
 
http://www.computeraudiophile.com/content/Audirvana-Decibel-Remoteapp
 
 
For Audirvana, I noticed that it was switching the bit rate to 24 automatically.  I thought this was the reason for the improvement I heard but if I just use iTunes in 24 bit mode, it actually sounds harsh and fatiguing.  Audirvana just sounds amazing to me.

 

[Yoga Flame]

Quote:

Are these differences in sound quality provable?  

[...]
 

One way to prove the differences is to record the output from the DAC. That is, connect the DAC's line out directly into a recording device's line in. By comparing the visual representation of the recordings, we should be able to see exactly where any differences are.

 

[DavidMahler]

Quote:

One way to prove the differences is to record the output from the DAC. That is, connect the DAC's line out directly into a recording device's line in. By comparing the visual representation of the recordings, we should be able to see exactly where any differences are.

This is a fantastic idea.  Someone want to do this?

 

[V-DiV]

Quote:

One way to prove the differences is to record the output from the DAC. That is, connect the DAC's line out directly into a recording device's line in. By comparing the visual representation of the recordings, we should be able to see exactly where any differences are.

Take one computer's (DAC's) audio output as input to another computer that has a program to record streaming digital audio, and then import the files into an audio editing program and compare.
 

 

[middachten]

Thats exactly what I did!
 
In fact I started to doubt the bit-perfect theory after being able to hear differences between these players. Other contributors to this topic are right: "there is no jitter inside the computer". So the only way that differences occur are because the software is not 'bit-perfect'.
 
I've buildt the following test setup: Macbook > TC-electronic Konnekt D24 SPDIF out > SPDIF in > Macbook. 
 
As test material I created an audio file with Audacity that was completely silent except for 1 sample going to maximum value. I used a 24bit/96kHz WAV format. This file looks like this:
 

 
Then I played AND recorded it with Audacity. It resulted in exactly the same puls shape. Conclusion: Audacity play/record cycle is bit perfect.
 
Now I went on and tested it with various players. During my first measurement cycle only the first measurement with Decibel was bit-perfect. My second measurement was VLC resulting in the following shape:
 

 
Vox, Fidelia and iTunes showed more or less the same shape. For people familiar with audio measurements the following picture of the same waveform (now with linear scale) will be more familiar:

 
This is what a FIR digital filter response looks like!
 
The next day, I wanted to continue with measurements. Then I discovered that Fidelia also gave a bit-perfect measurement!!!
But as soon as I had opened VLC or iTunes the original shape (see second picture) came back! Just to check if I was just creating ghost with my measurements I tested the dither settings in Fidelia. And they confirmed that my measurement setup did actually measure the player differences. The following pictures show 16bit and 24bit dither:

 

 
 
So, how come that the measurements are so different?
My theory is that as soon as VLC or iTunes are playing, the kernal/AU is using a different digital audio path.
When I closed VLC and/or iTunes and played a piece with Fidelia or Decibel in 'exclusive mode' the system was sort of Reset again. When i played the test file they both provided bit-perfect output.
It is not possible in this setup to test the 'exclusive mode' of these players because Audacity cannot record the input stream anymore (hence 'exclusive' 

 )
 
Anyone with an audio interface with SPDIF in AND out can verify this test.
If you need the test file please PM me and I'll mail it.
 
I'm very curious to hear what you think of these measurements.
Have I forgotten something?
Is there something not correct in my setup?
I would be pleased to get your input!

 

[turimbar1]

that impulse response with vox itunes and fidelia is pretty impressive, i am curious what upsampling would look like in this procedure (I cant imagine it looking too different), it would also be cool to see a log sweep of the audible freq spectrum.
 
very nice test, thank you middachten

 

[kwkarth]

Quote:

One way to prove the differences is to record the output from the DAC. That is, connect the DAC's line out directly into a recording device's line in. By comparing the visual representation of the recordings, we should be able to see exactly where any differences are.

With a complex audio signal, you'll hear differences before you see any.  Too much going on real time.
 
An impulse response test is not music.  Do a square wave test at 30Hz, 300Hz, and 3000Hz, and see what happens.

 

[middachten]

Quote:

An impulse response test is not music.  Do a square wave test at 30Hz, 300Hz, and 3000Hz, and see what happens.

I just wanted to check the 'bit perfect' hypothesis.
The only thing my measurements show is that in this setup the mentioned players are not (always) bit perfect.
At least Decibel and Fidelia CAN be bit-perfect.

 

[paaj]

I've tried the Amarra ('full') demo and Decibel and compared them to iTunes. I didn't hear a difference with Amarra, but in Decibel, the memory-mode seemed to put my system into 'warm' mode. Now I'm getting confused, the difference in sound can't be that big can it? First I thought it to be a little better: smoother, warmer, less bright.
 
But then I remembered I did hear this 'warm' sound in iTunes too sometimes, but since it only rarely pops up (1 in ~20 times I start my Mini) I thought it erroneous. Re-clicking the 24-bit output in AudioMidi (or 16-bit, same thing for CD anyway) and restarting iTunes always reset it to a sound I can replicate. I can't deliberately make it have the warmer sound. This is why I 'think' the brighter sound is the way it's meant to be, but now with Decibel the memory-mode makes the same kind of sound and it is supposed to be the best possible output right?
Very confusing stuff.
 
I thought I had no way of checking which signal is right, but didn't think about my Macbook as digital recording device. Maybe with your method I can capture the sound and see what is 'right'.
 
btw:
I use the digital out from the MacMini to my speaker system but the same warmer sound is also transmitted to my Airport Express for headphone listening, which implies it is not a DAC/optical-out fault but a software problem.

 

[holland]

That's a nice brickwall filter on the output.  It looks like you'd need 2 computers, one for play and another for record.
 
You can also try turning off sound effects in the sound set up, in hopes that it won't open an audio connection.
 
Playback and record a music sample.  Diff the files, I think Audacity can do that if you invert a sample.

 

[kwkarth]

Quote:

I just wanted to check the 'bit perfect' hypothesis.
The only thing my measurements show is that in this setup the mentioned players are not (always) bit perfect.
At least Decibel and Fidelia CAN be bit-perfect.

If I understood you correctly, you said you're taking the analog output from the DAC and re-digitizing it, and comparing?  Or did I get that part wrong?  At the very least you are converting to analog somewhere and then doing an ADC in and digital FFT and spectrum analysis.   So what's really happening?  If I did happen to understand you correctly, you will never know if you have true bit perfect or not because you're not even looking at the "bits."  You need to use a bit file comparator and compare the data files directly, before and after processing by the various players but before the data goes through a DAC..
 

 

[paaj]

I think he records the digital input on a macbook, so no ADC or DAC stages involved.
 
 
according to apple it's possible:
 
Quote:

connect your MacBook to a line-level microphone or optical digital audio equipment. The Audio In port accommodates both optical digital audio input and analog audio input.  
Analog line and optical digital audio input is accepted through a 3.5mm mini phone jack which does not provide power to a connected device, so you must use self-powered peripherals. The sound input jack accepts line-level stereo signals up to 24-bit stereo 44.1-192kHz sampling rate. It also accepts a stereo miniplug-to-RCA cable adapter for connecting stereo equipment to the computer.

Optical digital audio input is S/PDIF format and uses a standard Toslink cable with a Toslink mini-plug adapter, accepting up to 24-bit stereo and 44.1-96kHz sampling rate.