Sunday, December 30, 2007

Human Auditory System: Mosquito test your hearing

I've blogged before about the Mosquito ringtone - the teenagers' ringtone that adults mostly cannot hear.  Brian Dipert's excellent blog over at EDN (a technical magazine for the electronics industry) calls our attention to an in-depth article from last year in Sound and Vision Magazine on the Mosquito ringtone.  The article includes links to various MP3 recordings. Brian, who still has some high frequency hearing left even though, as I recall, he enjoys attending, as well as recording, live music performances that use sound reinforcement (i.e. loud, loudspeakers), is running a poll to see how many of us can hear the higher frequencies.

I missed the Sound and Vision article when it came out, but read it with interest today.  I do have some nits to pick with it though.  For starters, exposure to loud sounds is understood to be less harmful to young people than to us older folks.  The human ear has some automatic protection built in and although its effectiveness degrades as we age, it does enable younger people to be exposed to louder sounds without permanent damage.  Of course, there are limits to this ability, so this should not be taken by anyone as a license to go and blow their eardrums or cilia out(!).

My second nit is the author (David Ranada) slams higher sample rate audio formats.  His comments are certainly valid from the standpoint of continuous tones, but I believe that there are a good number of audiophiles and audio engineers who would argue that higher sample rates provide better reproduction of transient sounds.  This technical argument is based around the limitations of Fourier Transforms, which, very simply put, try to model all sounds as continuous sine waves.  There are also technical arguments for higher sampling rates based around the design of the anti-alias filters, which are easier and cleaner to implement at the higher rates.  Finally, I've heard that there has been some scientific research related to endorphin release in the brain - simply put, at low sample rates, we don't emotionally respond to digitally sampled music nearly as well as we do at high ones.

Now, I know that the whole topic of sampling rates is almost guaranteed to get most audiophiles riled, so let me close by saying that my opinion is not fixed on this point.  I was merely pointing out the arguments for the other side from a technical point of view. There are, of course, application, environmental, cost, convenience, and other factors that are involved in deciding what audio format to use and I have addressed none of those in this post.  That being said, I'll be happy to discuss them in the comments if anyone would care to.  I also have the best of intentions to write something on the use of high sampling rates in forensic audio and video processing in the near future.

Keith McElveen

P.S. For the record, Brian could hear the highest frequencies on his machine, but I could not, at least on my MacBook's speaker.


Herb said...

My problem with people pinning their hopes on higher-rate sampling formats is that the 16/44.1 digital formats we have now aren't the weak link in an audio system - the loudspeakers and the room are.

I assert that the typical listener will find a more dramatic improvement by treating their listening room and upgrading their speakers than upgrading to SACD/DVD-audio (mastering issues aside).

Of course there are real benefits of higher-rate sampling for recording (and, I would imagine, forensics).

Keith said...

Hi Herb,
I completely agree with you, particularly since you put the caveats about mastering, recording, and forensics in there.

To play off something you said in your comment, the recording aspect is one that I've been on my soap box about to my students - not so much in the controlled environment of a studio but instead in the often much less controlled environment of field recording. Having the extra dynamic range that comes with 24 bits (versus the standard 16 bits you pointed out) can ensure that you come back with signal and not just noise in a difficult recording situation. In noisy situations, bit depth (i.e. dynamic range) pays off much more than sample rate, in my opinion.