Tuesday, August 20, 2013

To listen to classical music you need a good pair of eyes


The periodical Nature reports on research by a social psychologist which shows that judgements about the quality of a musical performance are influenced more by what is see than by what is heard. The remit of the somewhat superficial research was the impact of body language, which means it did not consider a little-known but far more important link between the eyes and sound. It has long been a puzzle as to why high-order harmonics extending beyond the upper limit of human hearing produced by fine instruments such as Stradivarius violins make the music sound better. Similarly there has been no explanation as to why extending the frequency response of an audio system beyond the upper limit of human hearing improves the sound quality. But recent medical research has shown that our eyes are sound as well as vision transducers, and that the eyes play an important role in passing ultrasound to the brain. While the upper limit of human hearing ranges from 15 to 18 khz depending on age, the frequency response of the eye extends beyond 50 kHz - see graph above. In these ultrasonic regions the eye is not producing conventional sounds but is feeding sensory information to the brain which becomes a key part of the cognitive process.

The role of the eye as a sound transducer is medically proven, here is a link to research published in The International Tinnitus Journal. These findings open up many paths which this post can only hint at. Although top end audio systems have frequency responses that extends beyond 20 kHz, they come nowhere near matching the almost flat response to 50 kHz reported in the referenced article. Which may explain why even the best audio systems never quite seem to replicate the experience of live music. But when it comes to the ubiquitous compressed audio file formats such as MP3, the frequency response is further curtailed. Which may explain why classical music fails to connect with the MP3 generation. And that is before we factor in that headphones have become the default way of listening to music, and headphones remove the eyes completely from the listening process. While returning to live music, ultrasound is highly directional; which may explain why watching a performer closely seems to enhance the music.

Another fascinating possibility hinted at by this research is that John Cage's 4' 33'' is in fact an 'ultrasound symphony', with the absence of conventional musical sounds allowing the brain to focus on ambient ultrasound. And the concept of the eye as a multi-media transducer cross-references to a recent post on how cats can switch from one channel (hearing) to another processing track (sight). This action is scientifically described as synaesthesia and is the amalgamation of different sensory channels which usually function quite separately. Paths converge here as in the most common form of human synaesthesia sounds are perceived as images. So does the discovery that the eye is as an audio transducer explain why synaesthesia is common among musicians?

But most importantly - and I do think this is important - the Nyquist theorem, which is used to determine the sampling rate for digital audio formats, states that the maximum frequency that can be represented at any given sampling rate is half the sampling rate. Which is why CDs use a 44.1 kHz sampling rate, because that gives a frequency response extending beyond the limits of conventional hearing to 22.05 kHz. But research now shows that the brain responds to ultrasound beyond 50 kHz; so the data cut-off at 22.05 kHz may explain the perceived shortcomings of digital audio. And the absence of a 22.05 kHz cut-off in analogue LPs may explain why vinyl is making a comeback.

This extract from the conclusions to Martin Lenhardt's paper for The International Tinnitus Journal opens up a wealth of possibilities:

In regard to music recording and reproduction, more than doubling the sampling rate (95 kHz/24 bits) will extend the audible frequency range that can be coded in the eighth nerve and will result in a gain in linearity and reduction in quantizing errors, factors that will improve music quality.

Personal headphones could be supplemented or replaced with bone conduction transducers, with frequency responses extending to at least 50 kHz. Such transducers are already in use for medical treatment of tinnitus and can be readily modified for personal musical use (see Fig. 4).

Musical harmonic information is coded by place on the basilar membrane and temporally in neural firing. Ultrasound might contribute to the musical harmonic structure and provide more high-frequency treble emphasis in instruments, such as the cymbals, triangles, trumpets, violins, and oboes.
I came Martin Lenhardt's research paper while exploring the link between audio file format and sound quality. Inevitably my summary is simplistic, but further research on the role of ultrasound in music listening may help us understand why classical music is all too often lost in transmission.

* Part two of this post How classical music was covertly dumbed down is now available.

Also on Facebook and Twitter. Any copyrighted material on these pages is included as "fair use", for the purpose of study, review or critical analysis only, and will be removed at the request of copyright owner(s).

11 comments:

Lyle Sanford, RMT said...

I'd seen the Nature article - but had never heard of the Tinnitus Journal study, which does indeed open up a lot of paths.

Many thanks once again for your research and writing on audio/sound - the fundamental vehicle for music - that is so rarely discussed by musicians and those in the "classical" field gong on and on about increasing audience.

Pliable said...

Some hours after uploading this post I added a speculative reference to John Cage's 4' 33'' as an 'ultrasound symphony' and a more robust reference to the link between the eye as an audio transducer and synaesthesia. Those who read the earlier version of the post may want to revisit it.

MHF1953 said...

Sorry to puncture your theorizing, but you need to consider that the microphones cutting heads and amplifiers used to record the vinyl, as well as the pickup cartridges, amplifiers and speakers used to play it back are all just as limited, probably to even less than 20kHz in their bandwidth. The theory could certainly still apply to live performance, though.

Pliable said...

MHF 1953, most components in the audio reproduction chain handle signals well above 20 kHz as the following analysis of components in my main listening room show.

I have a graph in front of me which is the plot of the frequency response of the Technics AT-F3 moving coil cartridge that is in my Thorens TD125/SME Series 3 record deck. This shows the frequency response of the cartridge to be flat to 20 kHz. But because it was thought that frequencies above the upper limit of hearing are irrelevant, the graph does not continue above that point. Even if the frequency response drops away immediately above 20 kHz this still means the cartridge is operating well into the ultrasound region.

The Technics cartridge is amplified by an Arcam pre/power amplifier, which, as with most amplifiers, operates well into the ultrasound region.

The final link in the reproduction chain is a pair of Bowers & Wilkins 803 Nautilus speakers which are 6 dB down at 30 kHz and still working well above that point.

A similar situation applies to disc cutting heads. Quoted frequency response for later versions of the Neumann SX68 cutting head was 7 to 25000 Hz +/- 2 Db which means, again, operational range is well into ultrasound.

Don't be misled by the convention of audio equipment frequency responses being quoted up to 20 kHz. As above, this is because anything above was considered irrelevant, not because the performance falls off a cliff at 20 kHz.

Incidentally, as you are new to this personal website can I point out a couple of housekeeping matters. Identified rather than anonymous profiles are preferred - anonymous commenters usually get one shout and they are out - and smart one-liners such as "Sorry to puncture your theorizing" are not favoured.

MHF1953 said...

If offense was taken, it was not intended, and therefore, apologies are tendered.

On the frequency question, however: Yes, high end equipment is capable of passing signals well beyond the "normal" 20kHz cutoff. But unless we are speaking of relatively recent, rare, audiophile vinyl, the source material itself will roll off well before 20k. Once the high frequency is gone, it is gone, and no high bandwidth equipment will add it back. Having recorded my over 50 years worth of vinyl, 99% classical, I have rarely seen a record whose frequency output got as high as 20kHz (and I usually monitor the response in order to decide how much cleaning is required.)

Cordially,
M. H. Fischler

Pliable said...

MHF 1953, no problem. However in your original comment you said categorically “the pickup cartridges, amplifiers and speakers used to play it back are all just as limited, probably to even less than 20kHz in their bandwidth”.

But now you say “Yes, high end equipment is capable of passing signals well beyond the "normal" 20kHz cutoff”.

You also say assert “the source material itself will roll off well before 20k”. But again this is contradicted by empirical data - http://www.channld.com/vinylanalysis1.html (Note that “audiophile vinyl” usually have more care lavished on the recording and mastering process rather than access to unique wide bandwidth equipment).

Also other factors need to be taken into account, including the progressive roll-off of frequency response above 20 kHZ for analogue sources compared with the savage sampling rate induced 22.05 kHz cut off for digital sources.

This particular discussion has been useful but runs the risk of turning into a circular debate, so I am now amicably calling time on it.

There is a new related post here - http://www.overgrownpath.com/2013/08/how-classical-music-was-covertly-dumbed.html

Pliable said...

More on this path in Breaking - Norman Lebrecht gets it completely right - http://www.overgrownpath.com/2013/08/breaking-norman-lebrecht-gets-it.html

LasLab said...

Very interesting and I might add: "I knew it!" ;-) But one thing: it's a little conservative to classify music into classical and non-classical. All kinds of music, acoustical maybe more than electronic, will have this phenomenon.

Pliable said...

Thanks LasLab and you make a good point - we need to think outside the classical box.

Mariusz Herma said...

"Which may explain why even the best audio systems never quite seem to replicate the experience of live music. But when it comes to the ubiquitous compressed audio file formats such as MP3, the frequency response is further curtailed. Which may explain why classical music fails to connect with the MP3 generation."

Do you suggest that the power and beauty of classical music is more in what you can hear with your eyes than with your ears?

In the early and mid-20th century several generations all over the world fell in love with classical music by listening to poor transistor radio and cracky records and even worse players with the sound limited to ridiculous frequencies (by today's standards, even compared to current stream/mp3 quality). Eye response couldn't have had anything to do with this affair.

Pliable said...

"Do you suggest that the power and beauty of classical music is more in what you can hear with your eyes than with your ears?"

No, I am suggesting two things.

One is that we should open our minds to the possibility that new research on ultrasound sheds light on the differences between live and recorded music.

The other is that we should open our minds to the possibility that new research on ultrasound sheds light on long-standing concerns about the impact of digital encoding - with its abrupt frequency cut off - on music.

Yes, legacy music reproduction platforms had restricted frequency responses. But the restriction was in the form of a natural roll-off, not the abrupt cut imposed by 44.1 kHz sampling. The purpose of my posts is simply to suggest that we need to understand and, possibly, explore further little-known but important medical research.