Do files ripped from CD sound differet if they are FLAC vs MP3

[killdozzer]

Joking aside, you gain no benefit from sampling rates higher than 176khz, since 176khz contains all frequencies audible (and more) and captures transient information slightly beyond the limits of human perception.

A! There it is: high sampling rates and high frequencies!

 

[yepimonfire]

I still don't get the relations between digital information and reproduced sound. I do for some part, but then I get a strange feeling that certain measures are being used both to describe the workings in the digital domain and then later to describe the sound.

Like the other day, I saw a video where someone said something about a sampling frequency and then added how it is unnecessary as humans don't hear above 20kHz. I wanted to say; but you're not listening to the sampling rate, that's just the speed of "reading" a digital file, it seems as if you can't really have high audible frequency without a fast sampling rate (?).

EDIT: later confirmed by @yepimonfire

Then there's dynamics which imply both frequency span and dB and I hope in this case dB refers to those SPL dB's, otherwise I'm committing myself. This means that you need sufficient bit depth for music to be loud (or is it just in case it has both loud and silent parts?) and for music to go from low fq to high fq?


EDIT: later confirmed by @yepimonfire

Because everyone keeps saying we don't really need hi-res files and then comes Mark Walter in that lecture linked by @Auditor55 and says; sure, we don't need it until a certain instrument comes and hit a certain note (he was referring to a high pitch of some instrument). But, couldn't you then simply cut out the bottom part and still make it fit to a CD (which is not hi-res for him)

And then there's vinyl, which is said to have the equivalent of 8 bits, so it couldn't take a hi-res recording and yet someone raised this doubt just a couple of posts before. If it's known that vinyl is equivalent to 8 bits, and you need 16 for hi-res (or 24 for Walter), why raise this question? But you could cut digital info into record grooves, right? Like having a bump in the record groove for 1 and silence for 0, is it theoretically possible to record hi-res music onto vinyl this way? No matter how short the song might be, but could it be hi-res? Just make the "1 bump" more pronounced than record "clicks" to make it clear.

Sorry for trying your patience, I'm injured lying in bed, hence the intensified activity.

EDIT: I'm sorry, it's Waldrep. I never heard this surname and couldn't make it out from the video.

Humans cannot hear pure tones above an absolute max of 22khz, but we’re not talking about pure tones, but transient rise times. If someone strikes a cymbal, the rise time is in the microseconds range, which is still audible, 44.1khz cannot accurately capture that rise time, and you get temporal smearing. For those of you who have poor hearing above 16khz, set your sound settings in windows to output 32khz instead of
44.1. This will preserve all frequencies up to 16khz. Now listen carefully switching back and forth, you’ll notice the transients lose their definition and tightness, even though you can’t hear above 16khz.

An experiment was done awhile back where two old people in a nursing home, who had hearing issues and failed to hear anything above 10khz were played two samples of the same song, one was full bandwidth, and the other had a 12khz low pass. The couple could easily detect the low passed version. I’ll have to dig up the link to the study.


Sent from my iPhone using Tapatalk

 

[killdozzer]

you get temporal smearing.

Why do people then say we don't really need higher sq than RBCD? Is it just the ones who are not aware of this? Isn't this the best argument for the importance of inaudible parts of the spectrum? If this is the case, 320 mp3 isn't enough, no?

 

[mtrycrafts]

-30dBfs pink noise calibrated to 75dB leaves 30dB of headroom per channel, which is 105dB per channel, not combined. I generally watch movies at -15dB, which equates to 90dB per channel. During loud peaks with all channels blaring, I have recorded peak spl measurements of 110dB with the lfe going heavy in movies mixed hot. Each additional sound source adds 3dB, so all five channels at max equates to 102dB, add 10dB for the sub, and you get 112dB.

There is no real peak spl for music, since there’s no mixing standards. Classical music is about -30-20dBfs at average levels, while modern rock recordings are only about 6-12dB down.


Sent from my iPhone using Tapatalk

Yes, that is the max theoretical dynamic range possible but as with Cds, never ever get there.

 

[Steve81]

An experiment was done awhile back where two old people in a nursing home, who had hearing issues and failed to hear anything above 10khz were played two samples of the same song, one was full bandwidth, and the other had a 12khz low pass. The couple could easily detect the low passed version. I’ll have to dig up the link to the study.

Here's a different study:Audibility of a CD-Standard A/DA/A Loop Inserted into High-Resolution Audio Playback
http://www.aes.org/e-lib/browse.cfm?elib=14195

[Engineering Report] Claims both published and anecdotal are regularly made for audibly superior sound quality for two-channel audio encoded with longer word lengths and/or at higher sampling rates than the 16-bit/44.1-kHz CD standard. The authors report on a series of double-blind tests comparing the analog output of high-resolution players playing high-resolution recordings with the same signal passed through a 16-bit/44.1-kHz “bottleneck.” The tests were conducted for over a year using different systems and a variety of subjects. The systems included expensive professional monitors and one high-end system with electrostatic loudspeakers and expensive components and cables. The subjects included professional recording engineers, students in a university recording program, and dedicated audiophiles. The test results show that the CD-quality A/D/A loop was undetectable at normal-to-loud listening levels, by any of the subjects, on any of the playback systems. The noise of the CD-quality loop was audible only at very elevated levels.

 

[Irvrobinson]

Humans cannot hear pure tones above an absolute max of 22khz, but we’re not talking about pure tones, but transient rise times. If someone strikes a cymbal, the rise time is in the microseconds range, which is still audible, 44.1khz cannot accurately capture that rise time, and you get temporal smearing.

Rise time? The rise time of what? Yet again, you don't seem to have any idea how PCM encoding works, but you make these fantastic assertions that are bereft of technical basis. Temporal smearing? This sounds like more "inter-sample overs" baloney that seems fashionable in the recording industry lately. Care to explain this further?

[Hint: rise time refers to amplitude, not frequency. Sampling rate only affects the ability to accurately capture frequencies, not amplitudes. It's like you're talking about colors in terms of brightness.)

 

[yepimonfire]

Yes, that is the max theoretical dynamic range possible but as with Cds, never ever get there.

I can easily pull several soundtracks into a DAW right now and show several action movies do indeed reach that level in peaks. I’m not saying the entire dynamic range is utilized, just that the upper end at 0dBfs most certainly is. Same with cds too. Having stuff at 0dB doesn’t mean you need more dynamic range, the opposite in fact.


Sent from my iPhone using Tapatalk

 

[mtrycrafts]

I can easily pull several soundtracks into a DAW right now and show several action movies do indeed reach that level in peaks. I’m not saying the entire dynamic range is utilized, just that the upper end at 0dBfs most certainly is. Same with cds too. Having stuff at 0dB doesn’t mean you need more dynamic range, the opposite in fact.


Sent from my iPhone using Tapatalk

Dynamic range is defined as the difference between the loudest and softest tone recorded, not how loud one can play it. Not difficult to play Cds to 105 dB spl or louder.
Bit depth determines dynamic range possible not how the system is calibrated.

 

[Bucknekked]

Rise time? The rise time of what? Yet again, you don't seem to have any idea how PCM encoding works, but you make these fantastic assertions that are bereft of technical basis. Temporal smearing? This sounds like more "inter-sample overs" baloney that seems fashionable in the recording industry lately. Care to explain this further?

[Hint: rise time refers to amplitude, not frequency. Sampling rate only affects the ability to accurately capture frequencies, not amplitudes. It's like you're talking about colors in terms of brightness.)

Irvrobinson
I am glad you came to my rescue with your post. I wrote a rather lengthy but also rather unpleasant reply to the "temporal smearing" post. I erased it twice because it wasn't a helpful or illuminating reply. I don't have the technical chops to reply to "temporal smearing" but I know others have put that idea in its place with well supported viewpoints. Thanks for putting in a neutral and supportable reply.

 

[yepimonfire]

I still don't get the relations between digital information and reproduced sound. I do for some part, but then I get a strange feeling that certain measures are being used both to describe the workings in the digital domain and then later to describe the sound.

Like the other day, I saw a video where someone said something about a sampling frequency and then added how it is unnecessary as humans don't hear above 20kHz. I wanted to say; but you're not listening to the sampling rate, that's just the speed of "reading" a digital file, it seems as if you can't really have high audible frequency without a fast sampling rate (?).

EDIT: later confirmed by @yepimonfire

Then there's dynamics which imply both frequency span and dB and I hope in this case dB refers to those SPL dB's, otherwise I'm committing myself. This means that you need sufficient bit depth for music to be loud (or is it just in case it has both loud and silent parts?) and for music to go from low fq to high fq?


EDIT: later confirmed by @yepimonfire

Because everyone keeps saying we don't really need hi-res files and then comes Mark Walter in that lecture linked by @Auditor55 and says; sure, we don't need it until a certain instrument comes and hit a certain note (he was referring to a high pitch of some instrument). But, couldn't you then simply cut out the bottom part and still make it fit to a CD (which is not hi-res for him)

And then there's vinyl, which is said to have the equivalent of 8 bits, so it couldn't take a hi-res recording and yet someone raised this doubt just a couple of posts before. If it's known that vinyl is equivalent to 8 bits, and you need 16 for hi-res (or 24 for Walter), why raise this question? But you could cut digital info into record grooves, right? Like having a bump in the record groove for 1 and silence for 0, is it theoretically possible to record hi-res music onto vinyl this way? No matter how short the song might be, but could it be hi-res? Just make the "1 bump" more pronounced than record "clicks" to make it clear.

Sorry for trying your patience, I'm injured lying in bed, hence the intensified activity.

EDIT: I'm sorry, it's Waldrep. I never heard this surname and couldn't make it out from the video.

I actually greatly enjoy these “theoretical” tech discussions.

Bit depths higher than 16bit aren’t necessary for playback, only recording and processing (i.e. your receiver oversamples and increases the bit depth to 32bit/384khz before applying processing such as bass management, eq, Dolby Surround upmixing, time delay etc for a higher precision), oversampling also gets rid of the issues associated with the transform filter in the pass band. For playback purposes, bit depth only increases the range of the softest sound to the loudest sound. Sound below the maximum dynamic range, for example, in 16bit, which is -96dB, are lost due to quantization noise. In almost all cases, a dynamic range greater than 96dB is entirely unnecessary, let’s say you listen to a classical recording at an average level of 70dB, the average signal is -30dBfs, so the loudest peaks in the recording are 90dB, and the absolute quietest sounds that can possibly be captured are less than 1dB, i.e. inaudible. it’s highly unlikely that you will be able to even hear anything captured below 10dB due to the noise floor and auditory masking, even in an ultra quiet room. My own testing demonstrates that the quietest sound I can hear in a quiet room with a 30dBA noise floor using a 4khz sine wave is 1 dB, using full range pink noise, this falls to about 5dB. The 96dB dynamic range of 16bit is more than adequate. The only reason I mentioned needing 24bit for movies was because reference level is 105dB per channel, so a 5dB sound requires 100dB of dynamic range. Even so, you’d really only need 20bit, which is the equivalent bit depth resolution of the old Dolby Digital.

Still, if a film soundtrack were reduced to 16bit, you probably wouldn’t be able to tell any difference.

In the time domain, a 48khz sampling rate allows a maximum impulse of 20 microseconds, the rise time of a transient such as a cymbal strike can’t be properly recorded at this sampling rate, and humans can resolve a temporal difference of 6 microseconds. Yamaha has a good article about this. In addition, one study has shown the addition of ultrasonics to music elicits an emotional response http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0095464

You are correct that digital audio requires double the sample rate of the maximum frequency. Frequencies above half the sampling rate are not properly captured and fold down into the audio band as lower frequencies, which is why a very steep low pass must be used.

Although digital audio looks like a stair stepped waveform, it comes out as a continuous waveform after passing through the transform filter. To get an demonstrate how this works, you could create a 120hz square wave signal using rew and hook the output directly into the lowpassed input of your sub, rather than coming out as the irritating clicking sound of a square wave, it would sound like a relatively smooth bass sound, this is actually similar to how DSD works. The 60dB low pass used in the transform filter essentially removes the stair steps, since those stair steps are extremely high frequencies.

I have no idea what you’re getting at with recording a digital signal on vinyl. Due to the problems involved with the medium, I’d imagine the jitter during playback would be ridiculously high, as each successive playback wears the bumps down, 1s would eventually be interpreted as 0s.

Vinyl is an inferior format compared to hi res audio or even regular redbook. The only reason some vinyl releases sound better than digital is because it’s impossible to cut a record with a signal that has had a stupid amount of dynamic range compression applied, however, a digital release with less compression at 176khz would likely surpass the vinyl in every way.

Another interesting discussion is the DSD vs PCM debate. From a purely technical standpoint, I’d say DSD is worse than PCM. A good article discussing this viewpoint can be found here http://www.mojo-audio.com/blog/dsd-vs-pcm-myth-vs-truth/

The only way DSD would sound better than pcm is if it were played back on a 1bit delta sigma dac, since the conversion from pcm to pulse density modulation introduces noise and distortion into the signal (many cheap dacs such as those in a $30 DVD player use this type of dac), however, modern high quality oversampling dacs such as those found in our avrs don’t use this, it’s worth noting every study done using blind test comparisons between the two found no difference.


Sent from my iPhone using Tapatalk

 

[Irvrobinson]

Although digital audio looks like a stair stepped waveform, it comes out as a continuous waveform after passing through the transform filter. To get an demonstrate how this works, you could create a 120hz square wave signal using rew and hook the output directly into the lowpassed input of your sub, rather than coming out as the irritating clicking sound of a square wave, it would sound like a relatively smooth bass sound, this is actually similar to how DSD works. The 60dB low pass used in the transform filter essentially removes the stair steps, since those stair steps are extremely high frequencies.

What is a "transform filter"? Are you referring to the filter used to remove aliasing (junk samples for frequencies that exceed half the sampling frequency)? If that's what you're talking about, the aliasing filter does not "remove the stair steps". There are no "stair steps". The samples are used to perfectly reproduce the sine waves of the sampled analog input signals; there is no "smoothing" required. There is nothing to smooth. You've been looking at too many inaccurate graphics that claim to depict how digital audio works.

Why are you so intent on spreading around this sort of quasi-technical nonsense you've made up?

 

[Steve81]

There are no "stair steps". The samples are used to perfectly reproduce the sine waves of the sampled analog input signals; there is no "smoothing" required. There is nothing to smooth. You've been looking at too many inaccurate graphics that claim to depict how digital audio works.

Agreed. The basis of the Redbook CD standard, the Nyquist-Shannon sampling theorem, is pretty straightforward:

If a function x(t) contains no frequencies higher than B hertz, it is completely determined by giving its ordinates at a series of points spaced 1/(2B) seconds apart.

It's not a crude stair step representation. There's no caveat for rise time. The only issue is aliasing, which has been effectively solved for some time.

 

[killdozzer]

Although digital audio looks like a stair stepped waveform, it comes out as a continuous waveform after passing through the transform filter.

But I thought this is just a visual representation, a graph. I didn't think anything can be literally square shaped in sound reproduction.

I actually greatly enjoy these “theoretical” tech discussions.

Don't think you'll scare me away with this sentence! These quotation marks worry me. The term theoretical is used to describe just about everything from: "not true at all and I'm just throwing it out there" to "verifiable by hard data". I was hoping for the latter, but not all members seem to agree with you.

I'm interested in relations between digital domain and audible sound and haven't reached that chapter in "Sound Reproduction".

What makes it slightly more difficult is the fact that some terms are used in both, like frequency. As I said, someone skipped directly from frequency of sampling rates to what is audible. Which led me to believe that there's certain causality among them.

Hence:

Do you need high sampling rates to get a very high pitch/FREQ (not only what is enough, but ever)?

Do you need high sampling rates for music that goes from very quiet to very loud? (Think more than Private Investigations)?

Do you need high sampling rates for the music that goes from very low FREQ to very high FREQ?

And same goes for bits. I read more than once that since vinyl is equivalent to 8bits (10 the most whatever that means), it can never achieve the same dynamics as RBCD. This, at least to me, seems to imply that you need a certain amount of bits to achieve a certain amount of dynamics. So, again, is there a direct correlation?

Do dynamics always imply both huge oscillations in dB and FREQ or is one of these enough for certain material to be considered to have great dynamics?

And what if it's just timbre? (A lot of different instruments playing the same or similar tune, but same loudness and octave)

Is RBCD sometimes big enough to capture everything and sometimes not? Would you ever, ever need a hi-res audio for a Punk band?

 

[Bucknekked]

Bit depths higher than 16bit aren’t necessary for playback, only recording and processing

yepimonfire
if you stopped right there with the simple statement that greater than 16 bits aren't required for playback, only for recording and processing, I think you'd have most folks nodding their head in agreement and being generally agreeable. I can certainly subscribe to 16/44/1khz is enough to cover the audible range.

I am reading Floyd Toole's book, but haven't reached the detailed sections that cover this. But I know from his early comments in the book he subscribes to the RBCD standard as covering the audible range and can say that as an expert technologist, listener and other wise well regarded guru.

Agreed. The basis of the Redbook CD standard, the Nyquist-Shannon sampling theorem, is pretty straightforward:
It's not a crude stair step representation. There's no caveat for rise time. The only issue is aliasing, which has been effectively solved for some time.

Steve81
Thank you for this comment. I have heard the "stair step square wave" argument from many analog purists. Particularly guys like Neil Young and others who went out and started HD audio companies based on the flimsy idea that stair steps can't play like analog waves. Conflated ideas.

Don't think you'll scare me away with this sentence! These quotation marks worry me.

Is RBCD sometimes big enough to capture everything and sometimes not? Would you ever, ever need a hi-res audio for a Punk band?

Killdozzer
Keep those questions and the inquiring mind alive and coming. I learn something when someone asks an intelligent question and we get a variety of good replies.

I think your question about a Punk band and would they ever push the limits of the RBCD standard is exactly the issue. For most of the music we listen to, regardless of genre, do musicians push the limits of the standard with a bass, a drum kit, a guitar and a voice? What exactly does push the limits? And for how long?
I have heard classical enthusiasts weigh in that classical music, particularly a full orchestra playing a complex piece as being able to push the limits. Also, complex pipe organ pieces that stretch from dead quiety to thunder on the mountain.

But if the orchestral argument is true, and its also true that Vinyl can't hold as much information as RBCD can, they why is vinyl often the preferred format for many, many classic lovers? There is an obvious flaw in some of these statements. I know that an orchestra playing on vinyl can sound smashing. So if vinyl can't store as much info as RBCD, then the problem isn't related to how much can be stored.

I will admit that I don't have the technical chops to explain the issues. I am equipped to understand and follow a good explanation. I don't claim expert status here. Just interested newb.

 

[mtrycrafts]

..."temporal smearing" ....

Isn't that when you create two or more timelines?

 

[Bucknekked]

Isn't that when you create two or more timelines?

I think its when you get your shrimp tempora on the front of your shirt at a sushi restaurant

 

[yepimonfire]

Digital audio is stair steps until it goes through the near brick wall filter that removes frequencies above the nyquist. Once that happens it’s a smooth, continuous waveform, just like analog.


Sent from my iPhone using Tapatalk

 

[Irvrobinson]

Digital audio is stair steps until it goes through the near brick wall filter that removes frequencies above the nyquist. Once that happens it’s a smooth, continuous waveform, just like analog.

Are you learning impaired? Or are you just acting like a damned fool to keep this thread interesting? Those are the only two reasonable possibilities I can think of.

Edit - I forgot one. Are you stoned?

 

[yepimonfire]

yepimonfire
if you stopped right there with the simple statement that greater than 16 bits aren't required for playback, only for recording and processing, I think you'd have most folks nodding their head in agreement and being generally agreeable. I can certainly subscribe to 16/44/1khz is enough to cover the audible range.

I am reading Floyd Toole's book, but haven't reached the detailed sections that cover this. But I know from his early comments in the book he subscribes to the RBCD standard as covering the audible range and can say that as an expert technologist, listener and other wise well regarded guru.


Steve81
Thank you for this comment. I have heard the "stair step square wave" argument from many analog purists. Particularly guys like Neil Young and others who went out and started HD audio companies based on the flimsy idea that stair steps can't play like analog waves. Conflated ideas.

They misunderstand how digital works. After the low pass of the brick wall filter the wave is perfectly smooth. The issue I am referring to is impulse response. A sound cannot start and stop instantly in the pass band without containing ultrasonic frequencies, so you get ringing on transients. Obviously, speakers can’t stop and start instantly either, but ensuring the sample rate is high enough to capture an impulse within the limits of human hearing (6 microseconds) assures that the audio format isn’t a weak link in the playback chain. Those that brush of hi res due to the nyquist theory and the limits of human hearing in the frequency domain are completely ignoring the time domain, to which human hearing is much more sensitive.

But if the orchestral argument is true, and its also true that Vinyl can't hold as much information as RBCD can, they why is vinyl often the preferred format for many, many classic lovers? There is an obvious flaw in some of these statements. I know that an orchestra playing on vinyl can sound smashing. So if vinyl can't store as much info as RBCD, then the problem isn't related to how much can be stored.

I will admit that I don't have the technical chops to explain the issues. I am equipped to understand and follow a good explanation. I don't claim expert status here. Just interested newb.

Nostalgia. There is nothing about vinyl that is better than digital audio. Any sonic differences found in vinyl are nothing more than colorations, which are undesirable if accurate playback is the goal.




Sent from my iPhone using Tapatalk

 

[yepimonfire]

Are you learning impaired? Or are you just acting like a damned fool to keep this thread interesting? Those are the only two reasonable possibilities I can think of.

Edit - I forgot one. Are you stoned?

Do you understand how sampling and quantization works? I suggest you read the articles listed below before resulting to childish insults.

https://en.m.wikipedia.org/wiki/Digital-to-analog_converter

https://en.m.wikipedia.org/wiki/Pulse-code_modulation

http://www.yamahaproaudio.com/global/en/training_support/selftraining/audio_quality/chapter5/09_temporal_resolution/

http://www.yamahaproaudio.com/global/en/training_support/selftraining/audio_quality/chapter5/03_frequency_range/


http://emerald.tufts.edu/programs/mma/mrap/Batesweek6.doc


Sent from my iPhone using Tapatalk