Documentation Index
Fetch the complete documentation index at: https://docs.medera.info/llms.txt
Use this file to discover all available pages before exploring further.
VocalAcousticEngine extracts ~70 acoustic features using Librosa and Parselmouth (Praat).
Constructor
engine = VocalAcousticEngine(
sample_rate=16000,
frame_length_ms=25,
hop_length_ms=10,
gender=None, # 'male' (85–180 Hz), 'female' (165–255 Hz), or None (auto 75–300 Hz)
)
Primary method
features: VocalFeatures = engine.extract_features(audio: np.ndarray, sr: Optional[int])
VocalFeatures dataclass
F0 (fundamental frequency)
| Field | Unit |
|---|
f0_mean | Hz |
f0_std | Hz |
f0_min, f0_max, f0_range | Hz |
f0_confidence | 0–1 |
Voice quality
| Field | Unit | Normal range |
|---|
jitter | % | < 1.04 % |
shimmer | % | < 3.81 % |
hnr | dB | > 20 dB |
voice_quality_confidence | 0–1 | |
Prosody
| Field | Unit |
|---|
speaking_rate | syllables / sec |
articulation_rate | syllables / sec (excluding pauses) |
pause_count | int |
pause_duration_mean, pause_duration_total | seconds |
pitch_variability | coefficient of variation |
intensity_mean, intensity_std | dB |
prosody_confidence | 0–1 |
Spectral
| Field | Unit |
|---|
mfcc | 13-element list |
spectral_centroid | Hz |
spectral_bandwidth | Hz |
spectral_rolloff | Hz |
spectral_flatness | 0–1 |
spectral_confidence | 0–1 |
Clinical markers
| Field | Range |
|---|
vocal_depression_index | 0–1 |
vocal_anxiety_index | 0–1 |
vocal_distress_index | 0–1 |
prosodic_flatness | 0–1 (anhedonia marker) |
clinical_confidence | 0–1 |
Quality
| Field | Notes |
|---|
overall_quality | 0–100 |
voiced_fraction | 0–1 |
snr | dB |
audio_duration | seconds |
Library dependencies
- Librosa
- Parselmouth (Praat)
- SciPy (
scipy.signal, scipy.stats)
- NumPy
