Less change is more, except when it comes to money… part 2

Marshall Chasin
November 25, 2014

Last week, part 1 of this series examined Naomi Croghan’s research. She, along with her PhD supervisors and now colleagues, Kathryn Arehart and James Kates, had recently published an article in Ear and Hearing entitled “Music Preferences With Hearing Aids: Effects of Signal Properties, Compression Settings, and Listener Characteristics

They used a virtual hearing aid where music could be processed through 3 channels vs. 18 channels, and with a short release time (50 msec) and a long release time (1000 msec). Classical and rock music were piped through these various processing strategies and played to 18 experienced hearing aid users.

Last week, the effect of release time (fast at 50 msec and slow at 1000 msec) was discussed: Their research showed that the 1000-msec release time was perceived as being better than the shorter 50-msec release time. This can be summarized under the rule of thumb that “less change is more.” Improved perception of music (and undoubtedly speech) is obtained by either staying in a linear mode or staying in a WDRC mode for most of the time; switching back and forth may be problematic.

Along with studying the effects of release time on WDRC, these authors also looked at the number of frequency channels for the two types of music being studied. In this virtual (or simulated) hearing aid study, perception of classical and rock music was assessed in 18 experienced hearing aid wearers where the hearing aid was set for 3 channels and for 18 channels.

The study found, “For classical music… (the) number of channels was not significant. For rock music… three channels were preferred to 18 channels.”

This finding is rather counter-intuitive, at least for me. I wonder if they would have found the same results if there had been an option of only 1 channel. Here are my thoughts…

 

CLASSICAL VS. ROCK

Classical music has several salient differences from rock music: there are spectral differences, but, depending on the selection of the rock piece and the classical piece, this issue is highly variable and I would be unable to conclude how one spectrum might differ from the other. However, there is a major difference is in the size and number of string instrument players in a classical orchestra versus that of a typical rock band. A classical orchestra has violins, violas, cellos, and basses, and in total can easily amount to 30+ players. A rock band may have one violin as a “curiosity,” but it is certainly not “string-heavy”.

String musicians perceptively look for different things than a woodwind player, a guitarist, or a drummer.

When I, as a clarinet player, say, “That’s a great sound”, I am referring to the lower frequency inter-resonant breathiness of my clarinet. Even though my clarinet can generate significant high-frequency energy, I don’t need to hear it to make a judgement of the quality of my playing. A clarinet is a wonderful instrument for people with a (congenital or acquired) high-frequency hearing loss- I just need to hear up to 1000 Hz, and I’m fine.

In contrast, when a violinist says, “This is a great sound,” or we as listeners judge a certain violin as a wonderful sound versus that of a student model, we are making a judgement based on the relative magnitudes of the fundamental (or tonic, as musicians would say) and the higher frequency harmonics and overtones of the instrument. Hearing the violin at 4000 Hz and relating this sound to its fundamental at 262 Hz (middle C) gives us a judgment of music beauty. For string players, judgments of this sort are broadband, whereas for woodwinds they tend to be more restricted in required bandwidth.

Consider a 3-channel hearing aid (simulated or real) where the fundamental at middle C is at 262 Hz and the fifth harmonic is at 1572 Hz (6 x 262 Hz); the fundamental may be in the lowest frequency channel (perhaps 200-800 Hz); the fifth or so harmonic is in another separate channel (perhaps 800-2000 Hz); and the upper frequency components of the violin are in the top-most channel (perhaps 2000-7000 Hz).

In this scenario, the fundamental (or tonic) is compressed differently from that of the harmonic structure in channel 2, and also differently from the harmonics finding themselves in channel 3. In this multi-channel processing, the magnitude of the lower frequency fundamental at 262 Hz may end up having no relation to the magnitude of the higher frequency harmonics that are in different frequency channels. The important balance between frequency components would be lost.

Compare this now with a single 1-channel hearing aid. This is no longer commercially available, but it can be simulated on a virtual hearing aid. In this scenario, the relative magnitude of the lower frequency fundamental would be treated identically to the 10th or 20th harmonic and the balance would be maintained.

It could be that the processed classical music with “only” 3 channels was already significantly degraded (with anything more, such as 18 channels already being perceived as asymptotically poor quality).

 

FOR LISTENING TO MUSIC, LESS MAY BE MORE

Their finding that “For classical music… (the) number of channels was not significant. For rock music… three channels were preferred to 18 channels.” (p. e170) could perhaps have been more clearly stated as “for classical music anything greater than a single channel would be lousy.” I’m sure that they would not have said “lousy” but I would. Personally, I have listened to music through single channel hearing aids (such as the 1988 K-AMP from Etymotic Research) and they sound great for music, as opposed to more modern digital hearing aids, even when listened to at a quiet playing level.

I also wonder if, even for Rock music, a single channel would be better than the simulated 3 channel system? Again, we find that less may be more.

The hearing aid industry does need to create a single channel hearing aid for music, or at least a music program in the hearing aid that truly functions as a single-channel device. Multi-channel processing has major benefits for speech, especially in noise, but I believe that it was an error to extend this lesson from speech, to the processing of music.

Leave a Reply