V
Voice timbre & comfort analysis
Analyse the acoustic properties of recorded speech or any audio track: timbre, pitch, loudness dynamics, sibilance, and overall listening comfort.
Bash Markdown Python Text
Overview
This tool analyses audio recordings (or video tracks) to compute spectral features, pitch, loudness dynamics, sibilance, and timbre classification. It optionally uses Praat-based metrics for voice quality (HNR, jitter, shimmer, formants) and produces a heuristic attractiveness/comfort score along with visualisations (spectrogram, voice profile) and structured reports (text and JSON). The analysis pipeline includes automatic audio extraction from common video formats.
Tech stack
- Bash (wrapper script)
- Python (core analysis engine)
- librosa (spectral features, pitch estimation)
- praat‑parselmouth (voice‑quality metrics)
- numpy, matplotlib (computation and visualisations)
- pyloudnorm (perceptual loudness/LUFS)
- ffmpeg (audio decoding)
- FastAPI (optional advanced API service)