Welcome to emvoice’s documentation!
emvoice is a pure Python package for computing emotion expression-related features from speech signals. It uses similar algorithms as in Praat and openSMILE but also includes more recent methods from the librosa package. Currently, most low-level descriptor features from the extended Geneva Minimalistic Acousting Parameter Set (eGeMAPS) are implemented.
Given that it is entirely written in Python, it is easier to include emvoice in other Python-based applications, especially if they use numpy or scipy.
Installation
emvoice requires Python >=3.7 and can be installed via pip:
pip install emvoice
To install the lastet development version from GitHub repository, do:
git clone https://github.com/mexca/emvoice.git
cd emvoice
python -m pip install .
- Algorithms
- Loading speech signals
- Voice pitch
- Pitch harmonics
- Glottal pulses
- Jitter and shimmer
- Formant frequencies and amplitudes
- Harmonics-to-noise ratio (HNR)
- Root mean squared (RMS) energy
- Spectrogram
- Mel frequency cepstral coefficients (MFCC)
- Alpha ratio
- Hammarberg index
- Spectral slopes
- Spectral flux
- References
- Examples
- API Reference