Last Thursday, I had a chance to attend a talk on violin frequency analysis. The presentation took place in one of the lecture rooms at Solent University and was organised by the Audio Engineering Society South of England Group. Sara Rubio (BSc in Acoustic and Music and MSc in Human Responses in Combined Sound and Vibration Stimuli) presented the outcome of the last year experiment that she performed along with a luthier Jonathan Beecher. The idea was to see if there is a significant difference in harmonic content between mid-range "good" violins and extraordinary violins (Stradivarius, Gaudagnini).
The method involved three short recordings of a = 440Hz played on each of four violins. Sara used the same bow, volume and expression (last two as much as it's possible for human). The two mid-range "good" violins had their bridge tweaked by Jonathan to increase harmonic content. It was hard to draw a solid conclusion from the results.
As few people from the audience pointed out, the method used by Sara introduced to many variables. The make of wood was unknown, as well as the type and usage of strings. The data from FFT Spectrum analysis wasn't normalized and even for an experienced violinist, there is still +/- 3 dB tolerance while performing. Unfortunately, Sara and Jonathan had only one day to record and hence the spontaneous nature of the whole experiment.
Even though results brought more questions than answers I still had an opportunity to learn a lot. In the discussion afterwards, members of the Audio Engineering Society suggested using more promising impulse-response method, the same one used for convolution reverbs. With this method, we use short impulses to capture the reverberation of the acoustic space (in this case the body of the instrument). The data collected from such a technique can give us much more detailed information about the frequency response. Also narrowing down variables to the minimum is required if we want to draw more interesting conclusions. For now, all we can tell is that the better the quality of the violin the richer harmonic content we get, especially in the hi-mid part of the spectrum. Therefore it is suggested that a beginner should get a cheap instrument to reduce the annoyance of high pitched drilling tones...
There is no doubt that IR convolution is the future of many branches of the audio industry and not only. Game audio, cinemas, music, show business all of these and more will see improvements in immersive audio in coming years. Stay tuned.
Jake
Comments