Pitch shifting with the commercially available eventide eclipse: Intended and unintended changes to the speech signal

• Main Authors: Kolin, K.R (Author), Lupiani, A.A (Author), Murray, E.S.H (Author), Segina, R.K (Author), Stepp, C.E (Author)
• Format: Article
• Language: English
• Published: American Speech-Language-Hearing Association 2019
• Subjects: algorithm; Algorithms; analysis of variance; Analysis of Variance; female; Female; Healthy Volunteers; human; Humans; male; Male; normal human; physiology; pitch discrimination; Pitch Discrimination; speech; Speech; Speech Acoustics; young adult; Young Adult
• Online Access: View Fulltext in Publisher

 Purpose: This study details the intended and unintended consequences of pitch shifting with the commercially available Eventide Eclipse. Method: Ten vocally healthy participants (M = 22.0 years; 6 cisgender females, 4 cisgender males) produced a sustained /ɑ/, creating an input signal. This input signal was processed in near real time by the Eventide Eclipse to create an output signal that was either not shifted (0 cents), shifted +100 cents, or shifted −100 cents. Shifts occurred either throughout the entire vocalization or for a 200-ms period after vocal onset. Results: Input signals were compared to output signals to examine potential changes. Average pitch-shift magnitudes were within 1 cent of the intended pitch shift. Measured pitch-shift length for intended 200-ms shifts was between 5.9% and 21.7% less than expected, based on the portion of shift selected for measurement. The delay between input and output signals was an average of 11.1 ms. Trials shifted +100 cents had a longer delay than trials shifted −100 or 0 cents. The first 2 formants (F1, F2) shifted in the direction of the pitch shift, with F1 shifting 6.5% and F2 shifting 6.0%. Conclusions: The Eventide Eclipse is an accurate pitchshifting hardware that can be used to explore voice and vocal motor control. The pitch-shifting algorithm shifts all frequencies, resulting in a subsequent change in F1 and F2 during pitch-shifted trials. Researchers using this device should be mindful of stimuli selection to avoid confusion during data interpretation. © 2019 American Speech-Language-Hearing Association.