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: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica (Read 11677 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

What's with this new AES paper?

"The Effects of MP3 Compression on Perceived Emotional Characteristics in Musical Instruments"
http://www.aes.org/e-lib/browse.cfm?elib=18523
Kevin Graf :: aka Speedskater

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #1 –
What's with the AES? <- Rhetorical.

Somewhere along the way it lost its soul and now appears fully accepting of pseudoscience and other quackery.

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #3 –
 I wonder why that much research effort is done for topics that are pretty much outdated today. If somebody uses MP3, he can use VBR 0, CBR 320, or if he for some reason wants to be on the borders of audibility V1-V3. Therefore lower bitrates have only highly limited usage and their perceived character is a vastly speculative topic.

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #4 –
The AES only cares about science when it can be made to satisfy the commercial agenda, even if that means contorting it to such lengths that it no loner resembles science.

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #5 –
"Musical instrument sounds have distinct timbral and emotional characteristics"

Already in the first damn line of the damn paper, it's already 100% bullshit.

E: And the highest MP3 bitrate they used was 112kbps CBR. So the conclusion is that music sounds less good when highly compressed. What a momentous revelation!

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #6 –
Does this mean we can get command line tuning options to specify the emotional content of music when compressing? --angry  >:(

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #7 –
Ehhh...two pages in and apparently Spotify uses MP3 compression. Good to see their research didn't extend to Googling "spotify encoding."

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #8 –
Quote from: Rotareneg on
Does this mean we can get command line tuning options to specify the emotional content of music when compressing? --angry  >:(
With -b 32 --happy I will be happy with all 32kbps mp3s :)) . It looks like an article from the music faculty rather than the science/engineering department.

BTW the university in the paper is just several kilometers from my home.

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #9 –
Would it be a TOS8 violation if I said that the paper affected me emotionally, such that I felt an urge to compress it?

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #10 –
What do you think is wrong with their methodology?

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #11 –
Complete mystery on what are they trying to achieve,  next level could be amount of dizziness with Opus at 3kbps or what? It reads like they had to do a paper and they just did something out of thin air.
PANIC: CPU 1: Cache Error (unrecoverable - dcache data) Eframe = 0x90000000208cf3b8
NOTICE - cpu 0 didn't dump TLB, may be hung

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #12 –
Quote from: rw11 on
What do you think is wrong with their methodology?
In page 860 they showed the GUI of the test, asking user to rate which file is more "HAPPY", that means even if a user doesn't have any emotion against a test sample they are still required to choose one answer.

In page 863 the test result indicated that mp3 artifacts in 112kbps or lower are generally linked to negative emotions and I am totally not surprised with the results. Who likes compression artifacts?

A test forces users to choose an emotion of course will give results that compression artifacts have emotional characters!

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #13 –
Quote from: rw11 on
What do you think is wrong with their methodology?

The entire premise is flawed. They start out by stating that all instrument sounds have an emotional effect, but they show nothing to substantiate this spurious claim.

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #14 –
Quote from: rw11 on
What do you think is wrong with their methodology?
The authors tried to evaluate the impact of MP3 compression on emotional characteristics. And they discover that MP3 at low bitrate turns to really deteriorate it.
The big flaw of the analysis is that they are not really judging MP3 itself. By MP3 itself, I mean the MP3 algorithms and specific tools which induces specific artifacts or sound issue.

The authors indeed discovered that LAME MP3 at 32 kbps induces a lot of change in the emotional characteristic of music. But I recall that LAME@32kbps starts, before any compression, by resampling the signal to 16 KHz. Which means that before any kind of compression or MP3 algorithms, all frequencies higher than 8000 Hz are lost. Last but not least, LAME also lowpass the remaining signal to ~5800 Hz before —again— any compression.

In other words, the authors have discovered that an uncompressed PCM signal at 16 bit and 44100 Hz with a full spectrum of 0-22000Hz is more pleasant —emotionnaly wise— that a signal at 16bit 16000Hz and lowpassed to 5800Hz… And they are concluding that MP3 is responsible of the emotional loss.

A simple test:
• 20 seconds from a CD:
http://www77.zippyshare.com/v/slnlDg5b/file.html
• encoded to a ugly MP3 at 32 kbps:
http://www77.zippyshare.com/v/jgNRKNoS/file.html

ugly, isn't it.
• CD, resampled to 16KHz and lowpassed applied AND NO MP3 COMPRESSION:
http://www77.zippyshare.com/v/YymG1mfr/file.html

You can easily check by yourself that most of the sound degradation comes from the sound pre-processing and not by MP3 compression itself (which is rather close to the uncompressed signal at similar technical characteristic).


The main flaw is that they put the blame on MP3 instead of resampling/lowpassing.

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #15 –
Quote from: Rotareneg on
Does this mean we can get command line tuning options to specify the emotional content of music when compressing? --angry  >:(

Yes. You can pass a parameter varying from "utterly peaceful" to "aggggrrrressssive!".

In old days known as a "volume control". (Not perfect, but should have significant positive correlation. Unless you undo it in the playback process.)

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #16 –
Quote from: guruboolez on
The authors indeed discovered that LAME MP3 at 32 kbps induces a lot of change in the emotional characteristic of music. But I recall that LAME@32kbps starts, before any compression, by resampling the signal to 16 KHz. Which means that before any kind of compression or MP3 algorithms, all frequencies higher than 8000 Hz are lost. Last but not least, LAME also lowpass the remaining signal to ~5800 Hz before —again— any compression.
Well spotted, thanks! Ignoring, not to mention the nature of mp3 lowpassing and resampling the source with low bitrates is far from being a scientific attempt.
AES becomes a collection of papers in the spirit of An Enhanced Stereophile.
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #17 –
Allowing full bandwidth at 32kbps can actually degrade mp3 sound quality. I attached some fhg 32kbps files.

[1] using the default, which means allowing fhg to resample to 11025Hz and lowpass at 5175Hz
[2] my custom setting which forced fhg to use 44k sample rate and 22k lowpass.
[3] use SoX to resample to 11025Hz first and encode with default settings.

People who want to test LAME can do a similar test. If the current version doesn't allow full bandwidth and 44k sample rate at 32kbps, use an older version like 3.961 or lower.

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #18 –
This paper ranks high on the 'Comic' and "Sad" scales.   Even without reading it through one could have predicted the quality of the peer review here by the misuse , twice, of the word 'affected'  in the abstract.

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #19 –
Just as stupid as when a user at a industrial music forum i lurk. Went on how MP3 is never transparent & if it is its because the CD had poor mastering.  

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #20 –
Quote from: KozmoNaut on
Quote from: rw11 on
What do you think is wrong with their methodology?

The entire premise is flawed. They start out by stating that all instrument sounds have an emotional effect, but they show nothing to substantiate this spurious claim.

Right, and then they apparently tested using musical selections that are composed of the instrument playing steadily

"1.2 Instrument Sounds
We used eight sustained instrument sounds: bassoon (bs),
clarinet (cl), flute (fl), horn (hn), oboe (ob), saxophone (sx),
trumpet (tp), and violin (vn). The sustained instruments are
nearly harmonic, and the chosen sounds had fundamental
frequencies close to Eb4 (311.1 Hz)"

I'm depressed just thinking about it!

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #21 –
Quote from: Speedskater on
What's with this new AES paper?

"The Effects of MP3 Compression on Perceived Emotional Characteristics in Musical Instruments"
http://www.aes.org/e-lib/browse.cfm?elib=18523

Looks like it may be going viral:

compression-format-timbre-music-loss.html

Already more than  2 pages of direct hits on google...

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #22 –
Just as with the hi-res audio meta-analysis, it provides an easy and sensational (and wrong) headline to latch on to.

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #23 –
Could someone clever with statistics tell me whether they have allowed for multiple testing in the confidence limits- they are testing each bitrate against each of 10 emotions for each of 8 instruments. I would expect some results to turn up at 95% confidence level given that you have 80 goes at it per bit rate.

Give that at 112kbps on average *less than one* instrument showed significant results for each emotion, the result seems underwhelming.

I may have misunderstood though and am always happy to be corrected.

Re: The Effects of MP3 Compression on Perceived Emotional Characteristics in Musica

Reply #24 –
I was thinking about posting the paper in Hong Kong's audio discussion forums with a hope that the authors of the paper may visit or post in those forums, but stopped since I can imagine there will be some replies like...

"32-112k? 320k is already bad enough to let me commit suicide!"

"That's why DSD sounds heroic while PCM sounds timid."

"What? They use a Sound Blaster?"

...and so on.