Mora / Datasets / "speech recognition" / mostafaabdlhamed/speech-signal-features

speech recognition features Free and public

These features was extracted from large combine dataset which contain of these dataset (Ravdess, SAVEE, TESS, CREMA-D). These feature are consist of combing both prosody features like zero crossing rate and root mean square energy, and spectral features like mfccs. By applying these technique on each single audio, we have a large vector of features from single audio, then you can use these features to predict the emotions or personality and a lot of tasks like this. Here we divide a single audio into frames with fixed length and applying these feature extraction technique in each frame in audi