Spotify to launch 'Hi-Fi' CD Quality Tier.

[antcollinet]

Who said this test is reliable ?
What do we know about the files used in it ?

It would be an odd thing to fake. But there is a contact email on the site - why not contact them to ask.

 

[ads_cft222]

In practice, all streamed files music will be compressed. Losslessly or lossily, there's the difference.

Yes I replaced compressed with lossy

 

[antcollinet]

Just out of interest, I just tested myself on the 96kb/s lame MP3 vs lossless - I only got 76% correct.

Two of the tracks I was just guessing. I'd be guessing on all of them at 320. Probably even at 256.

(Speakers, not headphones)

 

[Leporello]

I have occasionally blind tested wav vs 128 kbit (converted by myself) with the Foobar abx comparator. Usually I have managed to get 20/20 results, but it has required concentrated listening. In most cases the difference is far from obvious. Could I enjoy listening the lossy version? Sure.

 

[Hayabusa]

I have occasionally blind tested wav vs 128 kbit (converted by myself) with the Foobar abx comparator. Usually I have managed to get 20/20 results, but it has required concentrated listening. In most cases the difference is far from obvious. Could I enjoy listening the lossy version? Sure.

Nice!
Did you try to up the bit-rate to see when you can not hear the difference anymore?

 

[antcollinet]

I have occasionally blind tested wav vs 128 kbit (converted by myself) with the Foobar abx comparator. Usually I have managed to get 20/20 results, but it has required concentrated listening. In most cases the difference is far from obvious. Could I enjoy listening the lossy version? Sure.

Which converter did you use? (All mp3 converters are equal, but some are more equal than others )

 

[Leporello]

Nice!
Did you try to up the bit-rate to see when you can not hear the difference anymore?

Only 320 kbit. No chance.

 

[Leporello]

Which converter did you use? (All mp3 converters are equal, but some are more equal than others )

I used Lame, supposedly the equalest of them all.

 

[ads_cft222]

Source : https://www.academia.edu/441306/Sub...ifferent_Musical_Genres?email_work_card=title

 

[sigbergaudio]

View attachment 365303



Source : https://www.academia.edu/441306/Sub...ifferent_Musical_Genres?email_work_card=title

A bit unclear text in that study. And unclear what point you are trying to make as well since you are just posting and highlighting the study with no comment.

The study says "These overall results demonstrate that mp3 compression does introduce audible artifacts."

While what it should have said based on their own study results is: "These overall results demonstrate that mp3 compression does introduce audible artifacts at 192kbps or lower, while at 256kbps or higher, the study has not proven this to be the case."

 

[Tell]

Just out of interest, I just tested myself on the 96kb/s lame MP3 vs lossless - I only got 76% correct.

Two of the tracks I was just guessing. I'd be guessing on all of them at 320. Probably even at 256.

(Speakers, not headphones)

To me the biggest difference with lower bitrate MP3 is that lowpass filter (which is at maybe 15khz at 128kbps afaik?), and if you're older than say 50 you won't hear over that frequency anyways. Don't know how old you are but I guess that might be the reason. Of course it's possible to hear other artifacts as well, but some samples are easier to encode than others.

But yeah, 320kbps OGG Vorbis is a totally different story, would be interesting to see a study on that and to have something to throw in the face of people claiming their wife hear a night and day difference from the kitchen when listening to an iPad in their car driving at 100kmh.

 

[Emlin]

To me the biggest difference with lower bitrate MP3 is that lowpass filter (which is at maybe 15khz at 128kbps afaik?), and if you're older than say 50 you won't hear over that frequency anyways. Don't know how old you are but I guess that might be the reason. Of course it's possible to hear other artifacts as well, but some samples are easier to encode than others.

But yeah, 320kbps OGG Vorbis is a totally different story, would be interesting to see a study on that and to have something to throw in the face of people claiming their wife hear a night and day difference from the kitchen when listening to an iPad in their car driving at 100kmh.

Driving their car in the kitchen? That's a new one!

 

[Tell]

Driving their car in the kitchen? That's a new one!

Yeah you know audiophiles living in enormous mansions with really large kitchens!

 

[antcollinet]

To me the biggest difference with lower bitrate MP3 is that lowpass filter (which is at maybe 15khz at 128kbps afaik?), and if you're older than say 50 you won't hear over that frequency anyways. Don't know how old you are but I guess that might be the reason. Of course it's possible to hear other artifacts as well, but some samples are easier to encode than others.

But yeah, 320kbps OGG Vorbis is a totally different story, would be interesting to see a study on that and to have something to throw in the face of people claiming their wife hear a night and day difference from the kitchen when listening to an iPad in their car driving at 100kmh.

62 and I can hear test tones up to about 13.5Khz.

I was not aware though that lower bitrate MP3 implemented low pass filters below 20Khz.

 

[Tell]

62 and I can hear test tones up to about 13.5Khz.

I was not aware though that lower bitrate MP3 implemented low pass filters below 20Khz.

Yup it does: https://wiki.hydrogenaud.io/index.php?title=LAME#Technical_information
That's only for variable bitrate though, can't find info for constant bitrate but I don't think anyone uses that anymore anyways (except for 320kbps) since it's not as efficient as VBR. But yeah apparently at around 128kbps the LP filter is set to around 16.5khz so neither you or me will hear that anyways

 

[ads_cft222]

A bit unclear text in that study. And unclear what point you are trying to make as well since you are just posting and highlighting the study with no comment.

The study says "These overall results demonstrate that mp3 compression does introduce audible artifacts."

While what it should have said based on their own study results is: "These overall results demonstrate that mp3 compression does introduce audible artifacts at 192kbps or lower, while at 256kbps or higher, the study has not proven this to be the case."

The other highlighted stuff is that specific groups could identify differences between cd and mp3 according to another study that this study references. And in general it discusses audibility capacity among different groups of people especially those acquainted with the natural sounds of musical organs . My aim is to show the ambiguity that exists in this matter

 

[sigbergaudio]

The other highlighted stuff is that specific groups could identify differences between cd and mp3 according to another study that this study references

Do you have a reference/link to that study then?

 

[ads_cft222]

Do you have a reference/link to that study then?

It is highlighted in my attached photo . Sutherland study

Citation number 8 you can search in the original link

 

[sigbergaudio]

It is highlighted in my attached photo . Sutherland study

Citation number 8 you can search in the original link

Unpublished manuscript from 2007? Color me not convinced.

 

[danadam]

That's only for variable bitrate though, can't find info for constant bitrate

I think it's here in optimum_bandwidth() in lame.c:

    const band_pass_t freq_map[] = {
        {8, 2000},
        {16, 3700},
        {24, 3900},
        {32, 5500},
        {40, 7000},
        {48, 7500},
        {56, 10000},
        {64, 11000},
        {80, 13500},
        {96, 15100},
        {112, 15600},
        {128, 17000},
        {160, 17500},
        {192, 18600},
        {224, 19400},
        {256, 19700},
        {320, 20500}
    };

This defines the start of the stop-band. Later it is rounded up to one of the 32 band-buckets that lame is using. For example for 17000:

17000 * 2 / 44100 = 0.771
23 / 31 = 0.742
24 / 31 = 0.774

therefore stop-band starts at:
24 / 31 * 44100 / 2 = 17071

and transition-band starts at:
(24 - 0.75) / 31 * 44100 / 2 = 16538

To verify:

]$ lame -b 128 test.wav test.mp3
LAME 3.100 64bits (http://lame.sf.net)
Using polyphase lowpass filter, transition band: 16538 Hz - 17071 Hz
...