Voice Timbre & Comfort Analysis

Analyzes recorded speech or audio tracks for timbre, pitch, loudness, sibilance, and listening comfort.

Bash Markdown Python Text

Overview

This tool extracts acoustic features from speech or any audio track, including spectral properties, pitch, loudness dynamics, and sibilance. It classifies timbre (dark/balanced/bright), estimates voice quality using Praat metrics (HNR, jitter, shimmer, formants), and produces heuristic attractiveness/comfort and confidence scores. Outputs include spectrograms, voice profile visualizations, a detailed text report, and structured machine-readable metrics.

Tech stack

Python
Bash
librosa
praat-parselmouth
pyloudnorm
numpy
matplotlib
ffmpeg