Design of a Vocal Quality Ranking System Using the P.YIN Algorithm Based on Python 3

Abstract

In the modern era of the music industry and vocal education, consistent practice is crucial to improving voice quality. However, the vocal evaluation process in recording studio environments is often subjective and not fully supported by objective and measurable evaluation software. This study aims to design and implement a vocal quality rating system called UtaSong based on PYTHON 3. The system utilizes the P.YIN (Probabilistic YIN) algorithm as the core engine for fundamental frequency (F0) extraction, implemented on a Wearnes AW-NE38H Mini Computer. The testing method is carried out by calculating the pitch deviation between the user's vocal and the reference (MIDI) in cent units, which is then automatically converted into a scoring system and grading. The test results indicate that the P.YIN algorithm successfully detects pitch consistently. Acoustic testing proves that a quiet environment (average noise of 38 dB) produces a much more accurate evaluation compared to noisy conditions (65 dB), where background noise affects the pitch deviation rate. In terms of computational performance, the hardware is functionally operational at the minimum threshold, with RAM utilization reaching 87–90% and GPU 98–99%. Therefore, increasing the RAM capacity to at least 8 GB and using a Solid State Drive (SSD) is highly recommended to optimize latency and overall system performance.