MP3 (320 kbps) VS flac/wav

[Anonymous]

What if you use a flux capacitor?

Ah, finally, an interesting subject. I like the fact that the DOC got it to run on garbage. Now that , my friend, is re-cycling.

 

[krazy_olie]

Can you pass both waveforms through a 20kHz low-pass filter, and show those results as well?

THat's the point really, a 20khz LPF will effectively be an antialisaing filter but a lpf will be prefect unless you have a very high order filter. Remember the nyquist theorem works mathematically but when you bring in real engineering costs it doesn't work quite as well.

Edit: not actually sure if it counts as an anti-aliasing filter as that (if i remember my filter theory correctly) would be done on the orginal waveform before the signal is sampled.

 

[MajorFubar]

No the waveforms are not dithered or filtered; naturally, dithering will smooth the signal a great deal in both instances, but the fact still remains that the 96kHz version gets closer in the first place. Dithering is clever and effective to a degree, but it can never really put back missing information. Real audio is far more complex than a basic sine wave, which I used for example and simplicity.

I haven't 'joined the dots' as such (though I know what you mean): I created the sine waves in Sound Forge and zoomed-in to roughly a 1 millisecond sample of each. I did it just to show visually the raw difference between 44.1kHz and 96kHz, because pictures paint a thousand words.

 

[krazy_olie]

No the waveforms are not dithered or filtered; naturally, dithering will smooth the signal a great deal in both instances, but the fact still remains that the 96kHz version gets closer in the first place. Dithering is clever and effective to a degree, but it can never really put back missing information. Real audio is far more complex than a basic sine wave, which I used for example and simplicity. I haven't 'joined the dots' as such (though I know what you mean): I created the sine waves in Sound Forge and zoomed-in to roughly a 1 millisecond sample of each. I did it just to show visually the raw difference between 44.1kHz and 96kHz, because pictures paint a thousand words.

Well I think theoritically it could be done but the point you illustrated very clearly is that it is more difficult. When stuff is engineered you have various constraints and you can't use perfect filters and what not. If the source is easier to handle in the first place then in the real world you should get a better output

 

[MajorFubar]

Correct. Whether you can hear the difference...well that's subjective again. Dithering is like blurring the edge of diagonal 'jaggies' in a low-res picture: it's successful in so far as it smoothes the join of contrasting pixels to make the image more acceptable, but ultimately it's not as good as having the picture in a higher resolution to start with.

 

[Anonymous]

THat's the point really, a 20khz LPF will effectively be an antialisaing filter but a lpf will be prefect unless you have a very high order filter. Remember the nyquist theorem works mathematically but when you bring in real engineering costs it doesn't work quite as well.

When applied to a signal before quantization, an LPF acts as an anti-aliasing filter. You are correct that you need a very steep filter to prevent frequencies above the Nyquist frequency from folding back and causing distortion, which is why a higher sampling rate is much better: they can work with a cheaper filter and still produce less aliasing in the audible range.

A low-pass filter applied to a discrete signal works as an integrator (smoothing filter): it smoothes out the waveform and removes the high-frequency components introduced by the discretization. You don't need a very steep filter for that, because most of the additional noise will occur at multiples of the sampling frequency instead of near the baseband signal.

I haven't 'joined the dots' as such (though I know what you mean): I created the sine waves in Sound Forge and zoomed-in to roughly a 1 millisecond sample of each. I did it just to show visually the raw difference between 44.1kHz and 96kHz, because pictures paint a thousand words.

Ok, I understand. I guess it's not really SF's fault either to draw it like that, but still -- no audio equipment will ever be able to produce a waveform like that, you would need an industrial-strength signal generator.

 

[Anonymous]

com oan geiezze a brake its getting farcical and funny with all the tech heads.

THat's the point really, a 20khz LPF will effectively be an antialisaing filter but a lpf will be prefect unless you have a very high order filter. Remember the nyquist theorem works mathematically but when you bring in real engineering costs it doesn't work quite as well.

When applied to a signal before quantization, an LPF acts as an anti-aliasing filter. You are correct that you need a very steep filter to prevent frequencies above the Nyquist frequency from folding back and causing distortion, which is why a higher sampling rate is much better: they can work with a cheaper filter and still produce less aliasing in the audible range.

A low-pass filter applied to a discrete signal works as an integrator (smoothing filter): it smoothes out the waveform and removes the high-frequency components introduced by the discretization. You don't need a very steep filter for that, because most of the additional noise will occur at multiples of the sampling frequency instead of near the baseband signal.

I haven't 'joined the dots' as such (though I know what you mean): I created the sine waves in Sound Forge and zoomed-in to roughly a 1 millisecond sample of each. I did it just to show visually the raw difference between 44.1kHz and 96kHz, because pictures paint a thousand words.

Ok, I understand. I guess it's not really SF's fault either to draw it like that, but still -- no audio equipment will ever be able to produce a waveform like that, you would need an industrial-strength signal generator.

 

[Alec]

What if you use a flux capacitor?

Incest nearly ensues, and the clocktower is in grave danger, and nobody wants that.

 

[Lee H]

What if you use a flux capacitor?

Incest nearly ensues, and the clocktower is in grave danger, and nobody wants that.

Great Scott!

 

[steve_1979]

I've just tried comparing 320kbps MP3's to lossless files using my new AVI system and I couldn't tell any difference between them.

Out of the 60 odd tracks that I compared there might have been a couple that I may have just about been able to hear a very slight difference. However these tracks sounded so similar that I think I was probably just imagining it. When I relistened to these tracks in a blind AB test I couldn't reliably tell any difference.

For this test I used 4 albums that I consider to be of the very highest sound quality in terms of recording and mixing. They were ABBA, Alan Parsons, Michael Jackson, and Mozart (Amadeus soundtrack).

 

[steve_1979]

I was considering re-ripping all of my CD's to lossless just because I have plenty of storage space and file size isn't an issue. However after this test I don't think it's worth the time and hassle. I'll just stick with MP3's.

 

[iemslie]

I've never posted in one of these debate type threads before (although I love reading them, people get so angry it's funny), but thought what the heck, why not.

In my experience it's difficult to nigh on imposible to tell the difference between lossless and decent high bitrate lossy rips, unless you sit down and concentrate extremely hard, and even then it's a bit hit and miss.

But who the hell wants to sit concentrating on every note? Surely the point of music is for either relaxing and enjoying, or jumping about and having a lark.

Does it bother me that god didn't bess me with bat like hearing? Not one bit(rate).

Cheers

 

[Andrew Everard]

I've just tried comparing 320kbps MP3's to lossless files using my new AVI system and I couldn't tell any difference between them.

:silenced:

For this test I used 4 albums that I consider to be of the very highest sound quality in terms of recording and mixing. They were ABBA, Alan Parsons, Michael Jackson, and Mozart (Amadeus soundtrack).

:silenced: :silenced:

 

[steve_1979]

But who the hell wants to sit concentrating on every note? Surely the point of music is for either relaxing and enjoying, or jumping about and having a lark.

You've sumed it up perfectly there iemslie. I think too many people forget this part and it's the most important bit.

Listening to high bit rate music on a good quality hifi is a very enjoyable pass time, but it's still the music not the quality that's important to me. At work we often listen to rubbish quality YouTube music through tinny PC speakers and I still get just as much enjoyment as I do from listening to my hifi at home, it just doesn't sound as good.

 

[steve_1979]

:silenced:

Andrew. Silenced.

Now that's a first.

 

[Andrew Everard]

Andrew. Silenced.

Not so much silenced as unable to find a 'biting my tongue' emoticon.

At work we often listen to rubbish quality YouTube music through tinny PC speakers and I still get just as much enjoyment as I do from listening to my hifi at home

There, just did it again.

 

[steve_1979]

If your unable to enjoy music unless it's played through several thousand pounds worth of hifi equiptment then I feel very sorry for you.

Where I work we don't just enjoy listening to music through tinny PC speakers. When we're outside I also have a little battery powered 1.5" mono speaker that we plug into our MP3 players. That tiny little £15 speaker has given my friends and myself many hours of enjoyment listening to music.

 

[Andrew Everard]

If your unable to enjoy music unless it's played through several thousand pounds worth of hifi equiptment then I feel very sorry for you.

Irrelevant though my unable is in this context, I fear you miss my point. And for that reason, I'm out...

 

[garyw77]

Forget the flux capacitor.....

Just "Reverse the polarity of the neutron flow" ...

 

[Clare Newsome]

Well I'm currently enjoying a wide array of digital files - from ITunes downloads to Lossless rips - being streamed via a DAC to active speakers, which is suiting a sunny Saturday reading the papers and doing some work. Also enjoyed listening to music in the car earlier and on my iPod while walking in the sun.

But if I really want to take my listening to another level I will switch on the main system and crank up the volume and the quality. To me, it's worth it.

There is room for all sorts of systems, and you don't have to spend a fortune or listen to Lossless music to enjoy your favourite tunes. But neither should anyone ever criticise anyone for choosing to invest their time and money on building their ideal system (and music library) whatever form/format that may take.

I wish people would respect the fact that while we all have different brains and ears, there is no right and wrong about hi-fi, just opinions.

 

[steve_1979]

I wish people would respect the fact that while we all have different brains and ears, there is no right and wrong about hi-fi, just opinions.

If I came across as being rude earlier it was unintended.

 

[Clare Newsome]

No problem. I just get incredibly frustrated/depressed by threads like this that

a) include vast amounts of techno-twaddle that bear little/no relation to the highly subjective, emotive experience that is music-listening and

b) incredibly arrogant assumptions that just because a poster has had one experience that everyone else *must* hear the same.

There are room for as many (if not more) opinions than there are choice of systems, formats and music

PS After this morning's digital delights, I am currently listening to heavyweight, 45rpm vinyl engineered by Steve Hoffman. Anyone whom doesn't think it sounds awesome is an idiot ;-)

 

[TALON1973]

i think it can sometimes come down to also "if you havn't experienced it you don't get it " , ive pretty much gone through the entire spectrum my of music ...expcept for stupid high end.... i think i'll class as over 3000 grands worth of h-fi as hi enough.. lol

now i'm in a good level of listening via my pc / senn hd600/ mf-vdac and gs solo srg2 for my music tastes...... and cant fault it either ... but i can tell a dud quality track over betetr bit rated versions a mile off... also still pull my vinyl on as well for a listen when i feel like it

 

[Mr. Iceman]

No problem. I just get incredibly frustrated/depressed by threads like this that

a) include vast amounts of techno-twaddle that bear little/no relation to the highly subjective, emotive experience that is music-listening and

b) incredibly arrogant assumptions that just because a poster has had one experience that everyone else *must* hear the same.

There are room for as many (if not more) opinions than there are choice of systems, formats and music

PS After this morning's digital delights, I am currently listening to heavyweight, 45rpm vinyl engineered by Steve Hoffman. Anyone whom doesn't think it sounds awesome is an idiot ;-)

SPOT ON Clare! :cheer:

 

[dannycanham]

Possibly, what do you mean with 4 distinct samples per second? Unless for some reason you're dividing 44.1 Khz "sample rate", by 10 KHz "frequency"... That's a bit like dividing 8 by fish. Either way, I can't tell the difference.

Not really, though I do confess I made a typing error due to being distracted (by my kids), and for that I apologise: I meant samples per cycle not samples per second, but the basis of what I said still stands: 4 samples per cycle of a 10kHz sound is not very much.

And, the Nyquist-Shannon theorum states simply that you need twice the sampling rate than the highest frequency to get "perfect reconstruction of the original wave". I understood that above that, it makes no difference to the end result as the algorithm will yield the same (perfect result). ie the sound at a particular frequency will be exactly the same. If it isn't, then the theorum is wrong, and someone needs to go back to the drawing board.

The theorem is correct that you need to at least double the sampling frequency, because audio waves have a 'plus' and 'minus' value which one sample per cycle wouldn't capture. But just two samples per cycle most certainly don't give a 'perfect reconstruction'.

The below screenshot shows two 1/1000th of a second snippets of a 10kHz sine wave. The top one is sampled at 44.1kHz and the bottom one at 96kHz. As you can see, the top wave is not very good at all while the bottom one more accurately represents a truer shape of the wave.

What on earth makes you think that a signal is reconstructed using straight lines? You can recreate a perfectly curved sin wave using a single point at each maximum and a single point at each minimum. You recreate waves using the rate of change between multiple points and not the shortest distance between two points.