Identifikasi dan Klasifikasi Suara Vokal Menggunakan Metode Fast Fourier Transform

Abstrak

This research develops a vocal range identification and classification system using frequency spectrum analysis and Fast Fourier Transform (FFT) algorithm. The system addresses the need for an accessible vocal range identification tool for amateur singers and the general public without formal musical training. The proposed system combines real-time audio recording capabilities with audio file processing, implementing pitch identification through FFT analysis and windowing functions. The system features two input methods: real-time recording and audio file upload supporting various formats (MP3, WAV, FLAC, AAC, OGG, M4A). Using PyAudio for real-time processing and Librosa for file analysis, the system accurately identifies fundamental frequencies within the human vocal range (80-1100 Hz) and automatically classifies voice types (Bass, Baritone, Tenor, Alto, Mezzo-soprano, Soprano). Testing demonstrates effective frequency identification with pitch conversion accuracy ranging from 95.7% to 98.8% and voice type classification achieving 81.2% accuracy. The system provides an efficient solution for vocal range analysis with low computational complexity and real-time processing capabilities.