Jump to content

FLAC and Hi-resolution Audio discussion


Timmy Fox

Recommended Posts

3 hours ago, crashjant said:

Enough! Stop poking us V2 version! There is a V3 with a non-working Hi-Res! We are all waiting for the update with the corrected Hi-Res support for 7 Android! All. Nothing more.

To be fair, high-res audio is more of a marketing ploy and a myth. I mean, unless you can actually hear dog whistles there is very little reason to go above the standard 44.1kHz ...

Link to comment
Share on other sites

12 hours ago, Timmy Fox said:

To be fair, high-res audio is more of a marketing ploy and a myth. I mean, unless you can actually hear dog whistles there is very little reason to go above the standard 44.1kHz ...

Yup, tested it myself on a copy of Muse - The 2nd Law yesterday, HD FLAC vs 320k mp3 and I could not tell the difference on a pair of HD600 with an ODAC and o2 amp (and I have decent hearing), which is hardly super high end but still more than the vast majority are going to be using.  I wasn't even comparing it to a standard flac.  I have a pair of JHA Roxanne's coming today and was going to see if I could tell the difference with those but didn;t even bother in the end, just deleted the HD flac (which was a ridiculous 1.2gb).  I believe on SOME tracks you can tell the difference between 320k mp3 and a standard flac but not on most of them (and I just spent ages swapping my 320k mp3's for flacs and testing them) and I doubt you could tell the difference with a HD flac at all.  Most people probably make the mistake of not normalising the volume with replaygain to get the 2 tracks the same volume, without doing that you have no chance of trying to tell the difference anyway.  Their second mistake is probably using bad flacs or mp3's to compare (i.e. fake ones that are not the correct bitrate), use Spek to help make sure they are real first.

Link to comment
Share on other sites

2 hours ago, ewokuk said:

Yup, tested it myself on a copy of Muse - The 2nd Law yesterday, HD FLAC vs 320k mp3 and I could not tell the difference on a pair of HD600 with an ODAC and o2 amp (and I have decent hearing), which is hardly super high end but still more than the vast majority are going to be using.  I wasn't even comparing it to a standard flac.  I have a pair of JHA Roxanne's coming today and was going to see if I could tell the difference with those but didn;t even bother in the end, just deleted the HD flac (which was a ridiculous 1.2gb).  I believe on SOME tracks you can tell the difference between 320k mp3 and a standard flac but not on most of them (and I just spent ages swapping my 320k mp3's for flacs and testing them) and I doubt you could tell the difference with a HD flac at all.  Most people probably make the mistake of not normalising the volume with replaygain to get the 2 tracks the same volume, without doing that you have no chance of trying to tell the difference anyway.  Their second mistake is probably using bad flacs or mp3's to compare (i.e. fake ones that are not the correct bitrate), use Spek to help make sure they are real first.

Speak only for yourself, there is a huge difference between a proper flac container and mp3 compression, stop spreading bs to everyone if you can not tell the difference fine for you, listen to mp3's. Other people can tell the difference and will listen to non compressed audio in flac, waw or dsd containers. There is no mistake if someone can not hear the difference doesn't meant there isn't one. I personally hate what mp3, or any other lossy format does to music, i hear a big difference so only listen to flac or straight cd.

Link to comment
Share on other sites

33 minutes ago, lantian said:

Speak only for yourself, there is a huge difference between a proper flac container and mp3 compression, stop spreading bs to everyone if you can not tell the difference fine for you, listen to mp3's. Other people can tell the difference and will listen to non compressed audio in flac, waw or dsd containers. There is no mistake if someone can not hear the difference doesn't meant there isn't one. I personally hate what mp3, or any other lossy format does to music, i hear a big difference so only listen to flac or straight cd.

true, but i still recommend you to listen .wav instead of .flac beceause flac is a compressed version of .wav for 30% butt there isn't a big difference between them. as you said there is indeed a huge difference between these formats mp3/wav/flac but on the upperhand, if you have crappy speakers or headphones you won't reconise the difference to.

Link to comment
Share on other sites

17 minutes ago, akincakiner said:

true, but i still recommend you to listen .wav instead of .flac beceause flac is a compressed version of .wav for 30% butt there isn't a big difference between them. as you said there is indeed a huge difference between these formats mp3/wav/flac but on the upperhand, if you have crappy speakers or headphones you won't reconise the difference to.

Compressed yes, lossy no. It's like a zip file, you do not lose any information. But we are starting another thread here, best stay with the long awaited update.

Link to comment
Share on other sites

There are no bad flac  files only flac  files with dodgy sources. Also wav (LPCM) and pcm are not the same. Finally 24bit audio 96khz for playback is mostly useless as both the dynamic range and output frequency are "beyond spec" for a human. A 16bit 48khz properly mastered music is way better than a dodgy 24bit file. You should research some more. 

http://people.xiph.org/~xiphmont/demo/neil-young.html

Link to comment
Share on other sites

20 hours ago, akincakiner said:

true, but i still recommend you to listen .wav instead of .flac beceause flac is a compressed version of .wav for 30% butt there isn't a big difference between them. as you said there is indeed a huge difference between these formats mp3/wav/flac but on the upperhand, if you have crappy speakers or headphones you won't reconise the difference to.

 

21 hours ago, lantian said:

Speak only for yourself, there is a huge difference between a proper flac container and mp3 compression, stop spreading bs to everyone if you can not tell the difference fine for you, listen to mp3's. Other people can tell the difference and will listen to non compressed audio in flac, waw or dsd containers. There is no mistake if someone can not hear the difference doesn't meant there isn't one. I personally hate what mp3, or any other lossy format does to music, i hear a big difference so only listen to flac or straight cd.

First they don't understand what lossless means, then they don't understand the range of human hearing. No doubt they are using $10 apple earphones for these claims as well.  I give up.

keep playing your HD files, I'm sure your dog will love it, hes the only one that's gonna hear any difference.

Link to comment
Share on other sites

5 hours ago, akincakiner said:

true, but i still recommend you to listen .wav instead of .flac beceause flac is a compressed version of .wav for 30% butt there isn't a big difference between them. as you said there is indeed a huge difference between these formats mp3/wav/flac but on the upperhand, if you have crappy speakers or headphones you won't reconise the difference to.

The statement about flac  is false, flac. is only the container the audio itself is pcm, flac is used as a zip. archive for audio and is a lossless format that should have absolutely no difference to the original source once uncompressed. Wav is not the same it is lpcm(linear) vs pcm, but sound quality should be identical, provided there are no glitches while decoding.

Speaker quality and specially headphone makes a huge difference, spent the same i did on my phone on my iem's, isine10. Even with my soundmagic e80's that i previously had I could not listen to mp3 when flac is and option. Lossless just sounds so much more alive and musical there is way more timbre in lossless music. I would assume part of the problem why people say they don't hear a difference is they are used to compressed lossy music and have no other reference point. Mp3 was a good enough(but only good enough, not close to optimal) solution back in early 2000's when storage was expensive, quality loss was acceptable, not anymore there is no excuse to go any lower than 16bit 44.1khz flac or wav.

Link to comment
Share on other sites

22 hours ago, Timmy Fox said:

To be fair, high-res audio is more of a marketing ploy and a myth. I mean, unless you can actually hear dog whistles there is very little reason to go above the standard 44.1kHz ...

Regardless if you believe there is no difference between Hi Res audio and CD quality audio (I disagree) either way the audio output on HTC Hi Res 24bit 192k is better quality than the standard 16bit 48k. But you will need good headphones to hear the difference.

Mastering makes a huge difference too. Well mastered CD quality audio can sound better than Hi Res audio with poor mastering. Well mastered Hi Res audio can sound absolutely fantastic, but again is dependent on you equipment. Without top notch headphones and DAP / amp etc. you won't hear much difference.

A lot of vinyl rips aren't that great, due to poor recording equipment and/or mastering, and this doesn't do much for promoting Hi Res audio.

Also, CD quality lossless FLAC is much better than any lossy source.

Link to comment
Share on other sites

8 hours ago, akincakiner said:

true, but i still recommend you to listen .wav instead of .flac beceause flac is a compressed version of .wav for 30% butt there isn't a big difference between them. as you said there is indeed a huge difference between these formats mp3/wav/flac but on the upperhand, if you have crappy speakers or headphones you won't reconise the difference to.

No.

Flac is like a zip file, you can compress a wav into flac and then extract the identical wav file back from the flac file again just fine. The audio data that comes out of your headphones or speakers will be identical regardless if you use wav or flac.

1 hour ago, HiFiZ said:

Regardless if you believe there is no difference between Hi Res audio and CD quality audio (I disagree) either way the audio output on HTC Hi Res 24bit 192k is better quality than the standard 16bit 48k. But you will need good headphones to hear the difference.

You're telling me my Sennheiser HD800 are not good enough headphones to hear the difference?

Ok, let me ask you if you actually know what you are talking about.

First off, let's address 192kHz vs 48Khz. If you're familiar with the Nyqvist theorem it is a scientific theorem that proves you need a sample frequency that is double that of which you wish to reproduce. This means that with 192kHz audio you can accurately reproduce sound with frequencies up to 96kHz. With 48kHz you can reproduce up to 24kHz frequency sounds. For a reference, the limit of human hearing is reached at around 19-20kHz though it gets lesser with age. If you've ever heard a mosquito, they emit sounds at around 17kHz (and is thus close to the limit of the very brightest sound that your ears are physically able to perceive). A dog whistle, depending on type, emits sounds anywhere between 23 and 54kHz and are known to be impossible to hear by just about any human (the same way you can not see infrared not UV light). Thus proves that both 44.1 and 48kHz is, per the Nyqvist theorem, more than enough to accurately reproduce any sounds that fall within the audible range of human hearing. Any more is only necessary if you wish to reproduce sounds of dog whistles.

As for bit depth (16, 24, 32 etc.) 16 bit audio is enough to natively reproduce up to 96dB of dynamic range. That is the difference between the loudest and quietest sound. 96dB is about the difference between a concert and complete silence. And I don't mean as in a quiet room in the middle of the night, that's still around 10 dB or so (a whisper is ~20dB), I mean complete silence which is only really achievable in special extremely insulated anechoic chambers. This 96dB figure however is not a true limit; There is also a technique called dithering which is applied to (without any loss of quality) increase this to an effective 120dB of dynamic range. This is the difference the complete silence I already explained and a sticking your ear right next to a jet engine. 120dB is loud enough to give you permanent hearing damage in seconds. Many consumer headphones and speakers will likely break before you exceed these levels of loudness anyway. Thus, 16 bit audio is enough to playback music loud enough to either break your headphones/speakers or permanently damage your hearing in seconds (whichever comes first).

Link to comment
Share on other sites

You do realize that just because it is 44,1khz or 48khz or 192khz most of the benefits come from time resolution increase and not the high frequencies(though they do also play a part, still being discussed what effect they actually have on listening experience). Meaning with 96khz audio you get 2x the samples every second with 192khz you get around 4x the samples, which can be desirable in some cases and produce more lifelike sound. Not to mention you do not need to apply such steep filters and don't have to filter at such low frequencies and anti aliasing becomes less of an issue in general. Most of the energy is focused in the low spectrum anyway. while the benefits of going 24 bit and beyond are far less.

Link to comment
Share on other sites

40 minutes ago, Timmy Fox said:

No.

Flac is like a zip file, you can compress a wav into flac and then extract the identical wav file back from the flac file again just fine. The audio data that comes out of your headphones or speakers will be identical regardless if you use wav or flac.

You're telling me my Sennheiser HD800 are not good enough headphones to hear the difference?

Ok, let me ask you if you actually know what you are talking about.

First off, let's address 192kHz vs 48Khz. If you're familiar with the Nyqvist theorem it is a scientific theorem that proves you need a sample frequency that is double that of which you wish to reproduce. This means that with 192kHz audio you can accurately reproduce sound with frequencies up to 96kHz. With 48kHz you can reproduce up to 24kHz frequency sounds. For a reference, the limit of human hearing is reached at around 19-20kHz though it gets lesser with age. If you've ever heard a mosquito, they emit sounds at around 17kHz (and is thus close to the limit of the very brightest sound that your ears are physically able to perceive). A dog whistle, depending on type, emits sounds anywhere between 23 and 54kHz and are known to be impossible to hear by just about any human (the same way you can not see infrared not UV light). Thus proves that both 44.1 and 48kHz is, per the Nyqvist theorem, more than enough to accurately reproduce any sounds that fall within the audible range of human hearing. Any more is only necessary if you wish to reproduce sounds of dog whistles.

As for bit depth (16, 24, 32 etc.) 16 bit audio is enough to natively reproduce up to 96dB of dynamic range. That is the difference between the loudest and quietest sound. 96dB is about the difference between a concert and complete silence. And I don't mean as in a quiet room in the middle of the night, that's still around 10 dB or so (a whisper is ~20dB), I mean complete silence which is only really achievable in special extremely insulated anechoic chambers. This 96dB figure however is not a true limit; There is also a technique called dithering which is applied to (without any loss of quality) increase this to an effective 120dB of dynamic range. This is the difference the complete silence I already explained and a sticking your ear right next to a jet engine. 120dB is loud enough to give you permanent hearing damage in seconds. Many consumer headphones and speakers will likely break before you exceed these levels of loudness anyway. Thus, 16 bit audio is enough to playback music loud enough to either break your headphones/speakers or permanently damage your hearing in seconds (whichever comes first).

I get where your coming from, but what I am saying is that regardless of the frequencies, on the HTC 10 specifically, the Aqstic chip is simply a better DAC than the standard Qualcomm DAC. And I can also hear the difference when upscaling CD quality FLAC files, I've done A>B comparisons and 192 kHz sounds better than 44.1 kHz on the 24bit Aqstic. My headphones are Beyerdynamic T51, which are a pretty good match with the HTC 10 imo.

I trust my ears more than any arguments pro or contra. The difference is subtle, but it's noticable. I would describe Hi Res vs CD audio as being fuller, more detailed, with more clarity and texture. This possibly has more to do with the mastering than the actual frequency etc., as digital Hi Res audio is more likely to be from a modern remastered source, and the mastering techniques used for Hi Res audio deliberately allow for greater dynamic range.

Link to comment
Share on other sites

For all you audiophiles here's an interesting article on FLAC vs high bitrate MP3. This person created a fairly in-depth test and issued a public challenge where people would evaluate the provided test material. I don't think anyone in the challenge used a phone for their audio source, but several used headphones. 

http://archimago.blogspot.com/2013/02/high-bitrate-mp3-internet-blind-test.html

One of the most interesting items was the result of which test set sounded inferior. 

Link to comment
Share on other sites

3 hours ago, lantian said:

You do realize that just because it is 44,1khz or 48khz or 192khz most of the benefits come from time resolution increase and not the high frequencies(though they do also play a part, still being discussed what effect they actually have on listening experience). Meaning with 96khz audio you get 2x the samples every second with 192khz you get around 4x the samples, which can be desirable in some cases and produce more lifelike sound.

This is incorrect. More samples does mean more data but it does not equate to more detail. This does not work the same as, say, frames per second in a movie or something of the like.

Having more samples per second is effectively just more of the same. As per the Nyqvist theorem, 2 samples per reproduced frequency has been mathematically proven to be enough to reproduce the recorded sound in its entirety. Adding more samples is practically just adding more dots to fill in an already perfect sine wave. This is the basically adding extra redundant points to a mathematical graph when you already have the proper points required to make a perfect sketch of said graph.

Or with an oversimplification; Adding more samples (say 4X the samples instead of 2X) is practically the same as adding 5+5+3+3 = 16 instead of 10 + 6 = 16. It's adding redundant samples to fill in an already perfect sine wave.

Link to comment
Share on other sites

30 minutes ago, Timmy Fox said:

This is incorrect. More samples does mean more data but it does not equate to more detail. This does not work the same as, say, frames per second in a movie or something of the like.

Having more samples per second is effectively just more of the same. As per the Nyqvist theorem, 2 samples per reproduced frequency has been mathematically proven to be enough to reproduce the recorded sound in its entirety. Adding more samples is practically just adding more dots to fill in an already perfect sine wave. This is the basically adding extra redundant points to a mathematical graph when you already have the proper points required to make a perfect sketch of said graph.

Or with an oversimplification; Adding more samples (say 4X the samples instead of 2X) is practically the same as adding 5+5+3+3 = 16 instead of 10 + 6 = 16. It's adding redundant samples to fill in an already perfect sine wave.

Nyquist frequency — The Nyquist rate is defined differently from the Nyquist frequency, which is the frequency equal to half the sampling rate of a sampling system, and is not a property of a signal.

I think you are understanding the theorem wrong, it says nothing about how much is enough and is only a theory for that matter. Correct one so far, but it's meaning is simple you need 2x the sampling rate to avoid any aliasing errors.

Link to comment
Share on other sites

55 minutes ago, lantian said:

Nyquist frequency — The Nyquist rate is defined differently from the Nyquist frequency, which is the frequency equal to half the sampling rate of a sampling system, and is not a property of a signal.

I think you are understanding the theorem wrong, it says nothing about how much is enough and is only a theory for that matter. Correct one so far, but it's meaning is simple you need 2x the sampling rate to avoid any aliasing errors.

Oh pardon my misspelling. Harry Nyquist was born in Sweden as Harry Theodor Nyqvist. I'm not entirely sure why his last name is more commonly spelled as Nyquist and not Nyqvist but he may have changed it as he emigrated to the US. Either way here in Sweden the spelling with a v is the most common one and is an old/alternative spelling for the Swedish word "Kvist", meaning a piece of a wooden branch or a twig. Despite being spelled with a u, it is still pronounced with a hard v-sound due to said origin. A lot of Scandinavian last names have their origin in nature, actually.

Either way, that is essentially a simplified version of how I was thought back in Electronics and Computer Sciences class. His proof shows that the lowest sampling rate required to accurately reproduce any given frequency is one that's twice of said frequency. Going higher (so called oversampling) does have some use in certain situations due to certain factors but as far as digital audio reproduction goes, 44.1 or 48 kHz has been proven time after time to be plenty.

 

Link to comment
Share on other sites

1 hour ago, Timmy Fox said:

Oh pardon my misspelling. Harry Nyquist was born in Sweden as Harry Theodor Nyqvist. I'm not entirely sure why his last name is more commonly spelled as Nyquist and not Nyqvist but he may have changed it as he emigrated to the US. Either way here in Sweden the spelling with a v is the most common one and is an old/alternative spelling for the Swedish word "Kvist", meaning a piece of a wooden branch or a twig. Despite being spelled with a u, it is still pronounced with a hard v-sound due to said origin. A lot of Scandinavian last names have their origin in nature, actually.

Either way, that is essentially a simplified version of how I was thought back in Electronics and Computer Sciences class. His proof shows that the lowest sampling rate required to accurately reproduce any given frequency is one that's twice of said frequency. Going higher (so called oversampling) does have some use in certain situations due to certain factors but as far as digital audio reproduction goes, 44.1 or 48 kHz has been proven time after time to be plenty.

 

Well not to me

Link to comment
Share on other sites

  • 3 weeks later...
On 7/1/2017 at 2:06 AM, yanqiyu01@gmail.com said:

But I can hear the difference between 16bit and 24bit

More than likely that difference is because it's two different versions of the same track and not just a plain 16bit vs 24bit.
If you instead take the 24 bit file and downconvert it to 16bit and do a blind test to compare, I doubt you'd be able to hear the difference.

All the bit depth does is affect the amplitude range. 16 bit is plenty for playing at a loud rock concert at full volume. By the time you would need more than 16 bits you would already be at the point where you'd have turned the volume up so loud either the speakers would have already broken or your permanently damaged your ears (if you're not already completely deaf by that point), whichever would come first.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...