The voice-gender from neuralnoise

Voice Gender

Gender Recognition by Voice and Speech Analysis

Read the full article.

This project trains a computer program to identify a voice as male or female, based upon acoustic properties of the voice and speech. The model is trained on a dataset consisting of 3,168 recorded voice samples, collected from male and female speakers. The voice samples are pre-processed by acoustic analysis in R and then processed with artificial intelligence/machine learning algorithms to learn gender-specific traits for classifying the voice as male or female.

The best model achieves an accuracy of 100% on the training set and 89% on the test set.

Update: By narrowing the frequency range analyzed to 0hz-280hz (human vocal range), the best accuracy is boosted to 100%/99%.

The Dataset

Download the pre-processed dataset as a CSV file.

The CSV file contains the following fields:

"meanfreq","sd","median","Q25","Q75","IQR","skew","kurt","sp.ent","sfm","mode","centroid","meanfun","minfun","maxfun","meandom","mindom","maxdom","dfrange","modindx","label"

"label" corresponds to the gender classification of the sample. The remaining fields are acoustic properties, detailed below.

In R, you can load the dataset file data.bin directly as a data.frame with the command load('data.bin').

In addition to the pre-processed dataset, the raw voice samples used for training are included as .WAV files in a separate repository. The .WAV files are pre-processed in R to produce the above dataset.

Accuracy

The trained models have achieved the following accuracies (train/test):

Baseline Algorithm (always male)

50%/50%

Baseline Algorithm (simple frequency threshold)

61%/59%

Logistic Regression

72%/71%

Classification and Regression Tree (CART)

81%/78%

Random Forest

100%/87%

Generalized Boosted Tree Regression

91%/84%

XGBoost

100%/87%

XGBoost (Updated with frequency range 0hz-280hz)

100%/99%

Acoustic Properties Measured

The following acoustic properties of each voice are measured:

duration: length of signal
meanfreq: mean frequency (in kHz)
sd: standard deviation of frequency
median: median frequency (in kHz)
Q25: first quantile (in kHz)
Q75: third quantile (in kHz)
IQR: interquantile range (in kHz)
skew: skewness (see note in specprop description)
kurt: kurtosis (see note in specprop description)
sp.ent: spectral entropy
sfm: spectral flatness
mode: mode frequency
centroid: frequency centroid (see specprop)
peakf: peak frequency (frequency with highest energy)
meanfun: average of fundamental frequency measured across acoustic signal
minfun: minimum fundamental frequency measured across acoustic signal
maxfun: maximum fundamental frequency measured across acoustic signal
meandom: average of dominant frequency measured across acoustic signal
mindom: minimum of dominant frequency measured across acoustic signal
maxdom: maximum of dominant frequency measured across acoustic signal
dfrange: range of dominant frequency measured across acoustic signal
modindx: modulation index. Calculated as the accumulated absolute difference between adjacent measurements of fundamental frequencies divided by the frequency range

Classification and Regression Decision Tree

The following decision tree, produced by the CART model, provides a high-level overview of important properties of the voice samples that may determine a specific gender classification of male versus female.

After narrowing the frequency range to 0hz-280hz with a sound threshold of 15%, the accuracy is boosted to near perfect, and the following CART model is described. Mean fundamental frequency serves as a powerful indicator of voice gender, with a threshold of 140hz separating male from female classifications.