Blind Listening Evaluation of Steel String Acoustic Guitar Compensation Strategies
A double blind multisample intonation rating test was administered to 32 experienced guitar players/guitar builders to test perceived effectiveness of some common steel string acoustic guitar intonation compensation strategies. The test used a randomized complete block design where each treatment was a typical guitar intonation compensation strategy. Each subject completed two sequentially presented sessions. Subjects were asked to rate intonation accuracy following audition of prepared sound clips. Each clip contained a short sequence of notes recorded from steel string acoustic guitar with either perfect intonation or tuning modified to fit the intonation profile of one of three typical guitar intonation compensation strategies: straight saddle compensation, individual string saddle compensation, or individual string saddle and nut compensation. Subject ratings indicate that all compensation strategies tested were equally effective. Analysis of test results by ANOVA did not indicate significant perceived differences for either session (p=0.596, p=0.286). Results of follow-up t-tests comparing intonation ratings for perfect intonation and the compensation treatment associated with the highest intonation errors (straight saddle compensation) also showed that these two treatments were equally effective in both sessions (p=0.137, p=0.359). Results of follow-up Bayesian estimation analyses comparing these two treatments also indicated no discernable difference for either session (session 1 difference of means 95% HDI: -1.31, 0.472; session 2 difference of means 95% HDI: -0.819, 1.13). Subjects’ correlation between ratings and actual intonation accuracy was determined by comparing ratings to intonation errors for each compensation strategy using Spearman's rank correlation. Rating correlation varied greatly among subjects (-1 ≤ ρ ≤ 0.949). The two subjects with the best overall correlations each gave identical ratings to two different strategies. These correlations showed low correlation to subject age and to years of guitar playing experience. These analyses suggest that all compensation strategies tested are equally successful.