Musicians have enhanced frequency representation. Narrowband spectrograms were calculated over the entire response to produce time–frequency plots (1-ms resolution) for musician and nonmusician responses to audiovisual (A) and unimodal (B) speech. Lighter colors indicate greater amplitudes. Musicians have greater spectral energy over the duration of the response than controls, with this difference being most pronounced at F0 (100 Hz). In addition, there was significantly more spectral energy at 100 Hz in the responses to audiovisual in contrast to unimodal auditory stimuli.

Musicians have enhanced onset response timing and magnitude. (A) Grand average onset responses of the musicians and control subjects to the AV (Upper) and UA (Lower) speech stimuli. UV speech and music stimuli elicited little activity, as indicated by the gray traces. Prominent peaks of the onset response (V, A, γ, δ) are indicated. Wave δ latencies were earlier in musicians than in controls. (B) Mean wave δ latencies for musicians and controls are shown with error bars denoting ± SEM. Musicians had significantly earlier latencies than controls in both the UA and AV conditions. (C) Musician and control grand average responses to AV cello stimuli. Mean rectified mean amplitude values were calculated over 4–10 ms (shaded gray) to test whether musicians (red) had larger response magnitude early in the subcortical stream, before cortical excitation. (D) Rectified mean amplitudes over 4–10 ms of the AV cello response indicated larger onset responses in musicians than controls to music stimuli.

Enhanced frequency representation in musicians and correlation with musical practice. (A) Fast Fourier transform analysis of the entire FFR period (30–350 ms) shows that musicians have more robust F0 peak amplitudes to both unimodal and audiovisual speech stimuli. (B) The mean F0 peak amplitudes (±SEMs) were significantly larger in musicians than controls for both unimodal auditory and audiovisual stimuli. (C and D) Years of consistent musical practice (>3 days/week) over the past 10 years (x axis) are plotted against individual peak F0 amplitudes in the UA and AV speech condition (y axis). The number of years subjects consistently practiced music correlated highly with the strength of speech pitch encoding (reflected in the peak F0 amplitude) for both UA (r = 0.73, P = 0.001) and AV (r = 0.68, P < 0.01) stimuli.

Stimulus timelines and audiovisual grand averages. (A) Auditory and visual components of speech and music stimuli. Visual components were digitized from videos of a speaker uttering “da” and a musician bowing a G note on the cello. Acoustic onset for both speech and music occurred 350 ms after the first video frame and simultaneously with the release of consonant closure and onset of string vibration, respectively. Speech and music sounds were 350 ms in duration and similar to each other in envelope and spectral characteristics. (B) Grand average brainstem responses to audiovisual speech (Upper) and cello (Lower) stimuli. Group amplitude differences were assessed by using a sliding-window analysis procedure that resulted in rectified mean amplitude values over 1-ms bins for each subject. Bins with significant differences (t test, P < 0.05) are designated by bars over the waveforms for each stimulus type. Amplitude differences in the responses between musicians and controls are evident over the entire response waveforms, especially in the speech condition. UV speech and music stimuli elicited little activity, as indicated by the gray traces.