A Database for the Acoustic Study of Male Bel Canto [a]
Description
All acoustic analyses in this study were performed using the Praat software to ensure data accuracy and reproducibility. The sound samples used and their corresponding acoustic parameters are provided in the supplementary materials. For ease of data management and reference, all samples were assigned a unified coding system. Samples numbered 1-16 correspond to those directly cited in the main text of the paper. As the total number of samples involved in the analysis exceeds 30, the remaining samples are sequentially numbered starting from 17. Furthermore, all original audio files have been submitted as supplementary materials. A strict one-to-one correspondence has been established between the audio files and the sample codes in the dataset, ensuring an accurate match between the data and the audio. This facilitates cross-validation and in-depth analysis by readers, peers, and experts.
Files
Steps to reproduce
The recordings for this study were conducted in a standard acoustic studio, with an initial reverberation time controlled between 0.3 and 0.7 seconds to effectively minimize the influence of room reflections on spectral sampling, thereby ensuring recording quality and signal-to-noise ratio. The recording setup employed a Neumann U87 large-diaphragm condenser microphone, coupled with a Millennia HV-3C microphone preamplifier and an RME UFX III professional audio interface, integrated with Cubase digital audio workstation software. This configuration ensured a flat frequency response curve, maximizing the preservation of the original acoustic characteristics of the voice while avoiding equipment coloration. During recording, singers were positioned approximately 30 cm from the microphone, which was equipped with a metal pop filter placed 15 cm in front to suppress airflow noise and maintain sampling stability.To ensure experimental rigor and data reproducibility, sample selection strictly controlled for variables such as gender, vocal fach, pitch, vowel type, singing technique, and recording environment. Following voice acquisition, preliminary editing and processing were performed using digital audio workstation software. Based on subjective auditory perception,the author classified approximately 400 samples, with the classification verified through three rounds of comparative evaluation to ensure that chēng hóu samples were accurately categorized according to perceived phonatory characteristics.The voice samples were imported into Praat software in a standardized WAV format, with a sampling rate of 48 kHz and 24-bit resolution. Each sample lasted approximately 3.5 seconds. Linear Predictive Coding (LPC) analysis was first conducted, with spectral slices extracted every 100 milliseconds starting from the second second, yielding 4–6 slices per sample. Concurrently, spectral analysis based on the Fast Fourier Transform (FFT) was performed. LPC and FFT results were integrated for presentation , providing a comprehensive representation of the acoustic characteristics.
Institutions
- Ankang University