Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Lossless vs. Redbook tests? (Read 115455 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Lossless vs. Redbook tests?

Reply #150
That's basically what I was getting at yesterday.  If decoding falls apart (eg: Monkey's Audio Insane on a Pentium II), the audible differences won't be subtle.

Moreover, it wouldn't result in "flat" or "lifeless" sounding... 


I'm glad you brought this up, because it makes no sense to me how you could even alter a digital music file to make it sound flat and lifeless.  To put it another way, if I asked you to alter a music file to make it sound flat and lifeless (without EQ, or some sort of compressor), could you do it?  If so what would it take?

Lossless vs. Redbook tests?

Reply #151
I checked in to CA again from your link....argh,the level of lame-brainedness certainly seems to have held steady there, complete with some patented audiophool favorite tropes,  at least until Axon came to tell them a few things I'd already said -- namely, that they can't necessarily trust their perceptions, and that there's plenty of science behind THAT assertion, and therefore they've been jumping the gun a bit in terms of evidence.  I'm perfectly OK with him being good cop to my bad cop on that score, it covers all the bases.  But even then several of the CA doofs *didn't get what Axon was saying* -- they simply 'saw' what they *wanted* to see, not what he actually wrote
Which is pretty much what happened to my posts there too -- and arguably to your first one here.


(Btw, I have seen Axon play more of the bad cop than this..and I must say he's good at that, too.  But if you think I wasnt' gentle enough, you should see what would happen if Arny or JJ posted there......ooh, the carnage.    )


Fair enough.  You seem fully aware of how you come across, and if brash/confrontational is your choice in conduct, so be it.  I spose that's why you chose the alias that you did. 

BTW - feel free to have Arny and JJ come on over.  It's been a few days since I've gotten a good lashing and I kinda miss it. 

Lossless vs. Redbook tests?

Reply #152
BTW - feel free to have Arny and JJ come on over.  It's been a few days since I've gotten a good lashing and I kinda miss it. 

It's not you who merits one -- far from it.  I mean,  look at what the particularly dense sooowhat took from all of Axon's careful explanations of scientific method, the meaning of scientific proof, ABX tests, evidence versus claims, statistics, etc...(emphasis mine): 

Quote
Axon,
actually, you appear quite the opposite, more specifically, you seem to be one of the most sane person I've encountered on audiophile forums. I wish you the best of luck in helping others come around to your understanding/use of ABX/double-blind testing, and that includes those who over-reach with their use of ABX testing in support of their 'opinions'.

Absent universal truths, it's really all just opinions, some of which are more popularly held, to be sure.


(/facepalm)

Lossless vs. Redbook tests?

Reply #153
That's basically what I was getting at yesterday.  If decoding falls apart (eg: Monkey's Audio Insane on a Pentium II), the audible differences won't be subtle.

Moreover, it wouldn't result in "flat" or "lifeless" sounding... 


I'm glad you brought this up, because it makes no sense to me how you could even alter a digital music file to make it sound flat and lifeless.  To put it another way, if I asked you to alter a music file to make it sound flat and lifeless (without EQ, or some sort of compressor), could you do it?  If so what would it take?

Haha no. It needs digital processing - But most (all?) audiophools don't realise this. I've seen several places and shops (even some that I usually have big respect for) sell DIGITAL cables that ENHANCES the image/audio... OMG I think I'm gonna puke...
Can't wait for a HD-AAC encoder :P

Lossless vs. Redbook tests?

Reply #154
I've seen several places and shops (even some that I usually have big respect for) sell DIGITAL cables that ENHANCES the image/audio...
There's an interesting thread on the Gearslutz forum about audible and measurable differences between digital (SPDIF) cables in a DA/AD chain. It's a bit early to draw any conclusions but it seems that there's more to it than "bits are bits". My guess is that it'll turn out to be DAC sensitivity to the quality of its input. They seem to be willing to do some serious testing.

Lossless vs. Redbook tests?

Reply #155
We shouldn't mix up layers. On the logical layer one bit is one bit, and with todays systems this is a fact with the inverse probability of winning the lottery seven days in a row. Digital systems as computers employ an armada of physical layer protocols to guarantee that the logical layer is consistent.

S/PDIF (AES/EBU) are physical layer links which put a lot of burden on the receivers side, don't use check summing, retransmission, etcs... This is usually ok, because transmission rates are really very low. Excellent S/PDIF receivers and transceivers are commodity parts today. There isn't any difference to be expected for any (not totally messed up) cable below 5m. Any contrary measurement means your DAC's PLL is either broken or badly designed. Any cheap plastic Toslink or by the meter antenna cable should be fine.

I once squashed a Toslink cable in a door. The insulation peeled of and the transparent plastic fiber already turned white at the kink. Out of curiosity I ran some jitter measurements with Wavelab through a Lynx card. It made no difference at all wether I used the fkuctup cable or a fresh one. My DAC's manufacturer advertises total jitter immunity, but I think any decent DAC shouldn't have any problems.

Lossless vs. Redbook tests?

Reply #156
I've seen several places and shops (even some that I usually have big respect for) sell DIGITAL cables that ENHANCES the image/audio...
There's an interesting thread on the Gearslutz forum about audible and measurable differences between digital (SPDIF) cables in a DA/AD chain.

Indeed interresting... I.e. I didn't know that digital jitter sounded like this:
Quote
its strange, some cable jitter adds a nice "analog sound" blur to the highs sadly also to all other frequencies in the digital signal.

Quote
with electric transparent digital cable midrange transparency improoves like crazy, also most harmonics, more punchier louder sound.


Thanks for clearing that up

Can't wait for a HD-AAC encoder :P

Lossless vs. Redbook tests?

Reply #157
Gotta love Gearslutz. I almost wish I lived in the same "reality" that most of the GS posters do, where even the slightest adjustment can yield Earth-shattering differences and in which there's absolutely no limit to the kinds of things audio equipment can do. Almost...

Lossless vs. Redbook tests?

Reply #158
If anyone is still interested in seeing hard evidence, I just succesfully reproduced my test by using SPDIF transport between separate PCs instead of using Virtual Audio Cable on a single PC.

The test setup:

- The same 30 s source clip as before, but resampled to 48 kHz (one of the devices can output only 48 kHz bit perfectly to SPDIF)
- The player PC: A Fujitsu-Siemens 2.53 GHz Pentium 4 laptop from the year 2002. Sound device: built-in sound, optical out through a combo minijack, a 3.0 m minijack to toslink optical cable (the cheapest I could find a few years ago)
- The recorder PC: The same 2.4 GHz P4 desktop as before. Sound device: Terratec DMX 6Fire 24/96, Terratec ASIO driver v. 5.53.03.144
- The used programs were the same as in the earlier test. (foobar2000/kernel streaming and Wavelab)
- This time the lossless format was FLAC -8 (I didn't bother to switch to another program from my audio editor just for creating an ALAC file.)

The result: All four files contain identical PCM data (= the two source files and the two files I cutted from the recording).
Foobar's bit compare cannot find differences and invert mix paste in Audition produces silence (amplitude = "-inf dB")

Possibly this kind of setup is immune to possible jitter effects because the PCM data is saved back to file, but at least the test proves that SPDIF can be used for lossless recording.

I can provide the source and result files if requested, but I don't think those bit-to-bit identical audio clips would be useful for anything.

Lossless vs. Redbook tests?

Reply #159
Foobar's bit compare cannot find differences and invert mix paste in Audition produces silence (amplitude = "-inf dB")


Alex, some of the guys over on CA are now claiming that even if the output of a lossless file is bit perfect (which they seem to accept), that "the timing of the bits may be off."  If this were true, when you inverted the lossless signal and combined it with the WAV file in Adobe Audition, there would have been some residual signal left, right?

PS - does "-inf dB" mean negative infinite dB?

Lossless vs. Redbook tests?

Reply #160
What timing of the bits? FIFO - 441000 in, audio out. Doesn't matter if audio is delayed by 0,5 seconds or miliseconds, what matters is that clocks are matched. You can't encode 441000 file, and get 44101 out. I mean, you can, but it would include some DSP in the way.
Error 404; signature server not available.

Lossless vs. Redbook tests?

Reply #161
they are so wrong and seeking last "excuses".

Again, see my long essay above:

Take a Lossless compressed (preferable FLAC) song file, of 16 bit 44.1 kHz stereo wav container,
but contents is not normal stereo 16 bit 44.1 kHz music,
it should be in our example for proof,
5.1 DTS multichannel DTS music inside the stereo 16 bit 44.1 kHz wav packed as Lossless FLAC / format of your choice.


If anything in the setup or transmission of the digital signals,
--- I talk only of the digital chain/processing, until you are on the analogue side inside the DAC, which outputs in this case 5.1 analogue, --- 
should be broken or bad configured,
you get ugly noise on the analogue side.

If the digital chain is well setup,
you get as result on the analogue side:
pleasing 5.1/multichannel music.
Quality of the multichannel music depends then only on the quality of the analogue componets, quality of the analogue output of the DAC and the rest of the analogue amping etc tec.

the Digital side can be achived nowadays sice some years for "very low cost" pefectly !!!

Analogue quality or perfection depends, is way more difficult than digital, but has nothing to do with the digital chain.
The analogue quality depends primarily on the 2nd half of the DAC, the analogue stage, of course, the source of analogue signals.
If the DAC adds dirt to the analogue signal, it will be dirty.

But the digital process itself until the 1st half inside the DAC in the digital stage, that is simple for the technics.



Testing with 5.1 DTS multichannel 16 bit stereo 44.1 kHz wav/FLAC is so much fun and should be proof and test enough for any HiFi setup,
because on the analouge side of the music processing chain of the HiFi,
the result is only black or white, NOT grey,
because either the data gets changed/broken/somewhat affected -> analogue ugly noise, no music, clearly listenable hickups in the music,
eg. if only small fingerprint on CD surface (error correction of CD-player can correct/interpolate with normal stereo music, but not with these special packed DTS...)
or the data is not changed in any way, time correct enough, and the DAC can do its proper work -> result to music, no noise, no hickups.
(quality of the music is not of interest because of the nature of 5.1-DTS inside of stereo-CDDA-wav-16bit-44.1KHz)
(An amp with 5.1 DTS decoding capabilities and digital inputs is required, be it optical digital or RCA-digital connection.)

Lossless vs. Redbook tests?

Reply #162
My DAC's manufacturer advertises total jitter immunity, but I think any decent DAC shouldn't have any problems.
There will always be "some" jitter left, either from the external or the internal clock. The problem is that we don't really know how much jitter can be assumed to be inaudible (probably also depends on its spectrum). I agree that it's up to the DAC designer to fix the jitter problem, but if there are no specifications for jitter levels, he can't know when it's good enough.
Back to the topic: digital data transmission and processing can be lossless, whereas DA and AD conversion can not. If jitter is an issue in this case, analyzing the bits is useless. It should be checked if the (analog) output of the DAC is different for lossless and redbook playback.
The outcome could (QED) be that a "broken" DAC sounds better (or worse!) with uncompressed redbook audio

Lossless vs. Redbook tests?

Reply #163
But when audio reaches DAC, it is in format DAC understands, it is not compressed with FLAC. It is already decompressed. It doesn't matter which file format was input.
Error 404; signature server not available.

Lossless vs. Redbook tests?

Reply #164
The outcome could (QED) be that a "broken" DAC sounds better (or worse!) with uncompressed redbook audio


Rather QED that you still don't get it. In a digital system there usually does not physically exists a continuous stream from the flac data on persistent storage to the output. FLAC files are read in chunks of several kilobytes into RAM. From there (possibly differently sized) chunks are fetched by the CPU, decoded and again written into RAM. From there they are send via PCI (Express) to one or more buffers (FIFO) within your sound output device. The systems are designed such that the final buffer is always filled enough so that bytes can be pulled at the set bitrate without running empty. Therefore output jitter is disjunct to anything (processing, decoding, etc.) before the last FIFO buffer. The only source of jitter could be a bad quartz clock inside the sound card or bad analog circuit design of the latter. It's not different for Redbook audio where the laser's raw signal passes several fifo buffers before the signal is ready to be fed into the output DAC.

There will always be "some" jitter left, either from the external or the internal clock. The problem is that we don't really know how much jitter can be assumed to be inaudible (probably also depends on its spectrum). I agree that it's up to the DAC designer to fix the jitter problem, but if there are no specifications for jitter levels, he can't know when it's good enough.


You have had a lot of the Koolaid, haven't you? Of course this is well known and perfectly measurable. You just compare the contents of receive and send buffer and continuously increase induced jitter between sender and receiver until buffer contents don't match anymore. Repeat this with different types of jitter and intensities, classify the results and there you are. No vodoo required.

Lossless vs. Redbook tests?

Reply #165
Alex, some of the guys over on CA are now claiming that even if the output of a lossless file is bit perfect (which they seem to accept), that "the timing of the bits may be off."  If this were true, when you inverted the lossless signal and combined it with the WAV file in Adobe Audition, there would have been some residual signal left, right?

My understanding is that as long as the transmitted data is in the digital domain it can be transmitted losslessly. It doesn't matter if the played PCM data was earlier stored in a lossless file format. It is always decoded to PCM long before it reaches the audio hardware. Also the Wave and AIFF file formats need to be processed before the PCM data can be transmitted. They are container formats that can contain data in various encoding formats.

In other words, the possible changes in the outputted analog signal (after the DA conversion) that may be caused by inaccurate clock signals are not anyhow dependent on the possibly used lossless compression. The played uncompressed and losslesssly compressed files will show similar random minor differences in the analog audio signal. As well you will not get exactly the same analog output again if you play a single Wave or AIFF file more than once in a row.

Quote
PS - does "-inf dB" mean negative infinite dB?

It is just how Audition indicates the amplitude of a completely silent file. The measured amplitude can be e.g. -6 dB or -59 dB below the full scale (0 dBfs). After I did the "invert mix paste" operation there was no content and the amplitude was detected as "minus infinite".

Lossless vs. Redbook tests?

Reply #166
You just compare the contents of receive and send buffer and continuously increase induced jitter between sender and receiver until buffer contents don't match anymore. Repeat this with different types of jitter and intensities, classify the results and there you are.
Jitter has to be quite high before it causes bit errors so I'm not worried about that. Imperfect jitter rejection of the DAC can have an influence on the reconstruction of the analog signal in the DAC so that's where should be looked for differences IMO.

Are there any details about the setup where people claimed to hear a difference ?


Lossless vs. Redbook tests?

Reply #167
If you are just worrying about jitter below the threshold of bit errors, there is no need to continue. And whoever told you that this is relevant, rethink wether this source is trustworthy.

For an adequately designed DAC this has no relevancy, not even the slightest. If not, dump your DAC and anybody recommending it. To draw a line under this once and for all, read Robert W. Adams' article about jitter & DAC design in the 1994 issue No. 21 of the The Audio Critic.

Lossless vs. Redbook tests?

Reply #168
Foobar's bit compare cannot find differences and invert mix paste in Audition produces silence (amplitude = "-inf dB")


Alex, some of the guys over on CA are now claiming that even if the output of a lossless file is bit perfect (which they seem to accept), that "the timing of the bits may be off."  If this were true, when you inverted the lossless signal and combined it with the WAV file in Adobe Audition, there would have been some residual signal left, right?
No, this theoretical difference would be present in the analogue domain only.

As other have said, the FLAC decoding is far too far up the audio pipeline to have any bearing on this. Both FLAC and WAV go through multiple buffers after this - the timing of the FLAC decode is irrelevant.

If the other operations of the PC (e.g. decoding FLAC vs playing WAV, browsing the net, running the operating system(!), refreshing the display etc) change the timing of data as it finally reaches the DAC, then there's no hope.

It's not impossible - it's perfectly possible if something is broken - but it means audio playback will also fall apart when anything else happens on the PC - e.g. when the seek bar, spectrogram or volume indicator move in the playback program!


btw, the timing of bits (=jitter) is perfectly measurable to a far far greater accuracy than the human ear can resolve it, and is often measured. The result? "decent" sound cards don't have any of these issues - they don't care what the PC is doing (decoding FLAC, WAV, gaming, etc!) as long as the audio buffer doesn't empty.

Cheers,
David.

Lossless vs. Redbook tests?

Reply #169
It's not impossible - it's perfectly possible if something is broken - but it means audio playback will also fall apart when anything else happens on the PC - e.g. when the seek bar, spectrogram or volume indicator move in the playback program!

If something is broken it won't make any difference if you're playing uncompressed or compressed because...

As other have said, the FLAC decoding is far too far up the audio pipeline to have any bearing on this. Both FLAC and WAV go through multiple buffers after this - the timing of the FLAC decode is irrelevant.



Lossless vs. Redbook tests?

Reply #170
If something is broken, the load on the CPU or HDD (for example) could interact with the DAC clock or power supply. It doesn't matter how many buffers you put in between.

This is really unlikely, and would be relatively easy to measure. Depending on the nature of the fault, the FLAC might cause fewer problems than the WAV!


btw, and slightly OT, there's an interesting review of a $5000 music server in Hi-Fi News this month. The audio output is an M-audio 8 channel PCI sound card. The measurements are quite ordinary - i.e. easily measurable noise and hum (despite additional screening added to the sound card), but good jitter results.

Cheers,
David.

Lossless vs. Redbook tests?

Reply #171
Took the DTS-CD test and got bit perfect results . . .

It was mentioned a few posts back that DTS-CDs (which have 6 channels of surround sound encoded into a standard 16bit 44.1Khz stereo track) need to be reproduced with bit perfect output in order for them to work. If any part of the signal is corrupted, you get static instead of music.

Well, I bought two DTS surround encoded CDs off of Amazon and just ripped them into my Mac Mini using the Mini's CD drive, the iTunes ripper, in ALAC format. Surprise, surprise, the ALAC files of these two discs play just fine, even output via the built-in optical in the Mini. I've listened to both CD's in there entirety, and never once heard a hiccup. Well, I did hear about a second of static when I was loading some album artwork, but that doesn't really count.

I'm moving on . . .

Lossless vs. Redbook tests?

Reply #172
Alex, some of the guys over on CA are now claiming that even if the output of a lossless file is bit perfect (which they seem to accept), that "the timing of the bits may be off."


In the context of SP/DIF or TOSLINK connections the timing of the bits can be relevant. The DAC clock of a SP/DIF or TOSLINK  DAC  is based on the digital input. If sufficiently corrupted, the possibility for audible degradation exists.

A properly designed DAC will either produce good-sounding audio, or mute itself. 

The history of DACs is that lot of poorly-deisgned products made their way into the marketplace, particularly the high end audio market from about 1985 to maybe as late as Y2K.

For example, I tested two DACs, one a  Denon DA 500 toslink and SP/DIF DAC, the other a Panaonic SHAC 500 surround decoder. 

I devised a means for applying variable amounts of jitter to an audio signal.  By jittering the input to the Denon DA500 enough, I could make music sound clearly wattery, and produce tons of jitter in measurements.

I could jitter the input to the SHAC 500 far more with zero audible and measurable effect. At the limit the SHAC 500 would simply mute rather than sound bad or even measure to have even the slightest measurable degradation of test signals.

Lossless vs. Redbook tests?

Reply #173
btw, and slightly OT, there's an interesting review of a $5000 music server in Hi-Fi News this month. The audio output is an M-audio 8 channel PCI sound card. The measurements are quite ordinary - i.e. easily measurable noise and hum (despite additional screening added to the sound card), but good jitter results.


Reading between the lines the audio interface in question sounds like a Delta 1010LT, which is hardly a top-of-the-pile device.  Adding screening around any competently designed audio interface should be futile, but it might impress visiting firemen!

 

Lossless vs. Redbook tests?

Reply #174
As a longtime audio forum participator, a recent participator at Computer Audiophile, and sometime lurker over here I would like to commend Axon for his contributions over there. 

Most of the rest of you, on the other hand, deserve a big wag of the finger.  Because of the assumptions of many people here, it took the better part of a week before relevant explanations were made of why it is useless to try to test AIFF vs. lossless.

The stereotyping and prejudice on display in this thread, the questioning of whether it is even worth it to share knowledge, should cause you shame. 
Greynol's comment, coming from such a bigwig over here, was particularly disturbing.  "Consider yourselves lucky."  Just what is a forum for, anyway?  So you guys can put down people for being idiots and then refuse to share what you know?

What is really at issue is that most of you were unwilling to or incapable of communicating constructively.  You didn't figure out where the sticking points were for the people trying to understand, and you didn't offer effective explanation of those points.  Then you blame a handful of extremists and abandon the vast middle, unwilling to brook any confusion or failure to immediately accept your point of view.  True, many of you have points of view that accord with incontrovertible fact, but when you assert it without explaining it fully and convincingly, it is just another opinion to any rational non-expert.

Everybody over there pretty much accepted that the files were the same, but had questions about how the data is processed and whether that affected sound quality or jitter.  In general this reflects a general misunderstanding among audiophiles and newbies about when bits is bits and when there are timing issues (with arguable effects).  Not some general irrationality, but as OP says, just lack of knowledge.

Once the architecture of pc audio was explained, specifically the irrelevance of anything before the final buffer on the soundcard, the discussion popped into focus for me and for many others, I expect.

True, there is a big issue in that many people are unaware of the unreliability of hearing impressions and put too much stock in their own or in those of certain charismatic figures, and I understand it is frustrating, but on the other hand, look at the unquestioned assumptions in this thread (which Axon analyzed above).