hypersunflower/ava_speech_data_log_mel_spec

this dataset was created using data from nccratliri/vad-human-ava-speech

i just converted audio to log mel spectogram, as a part of my pet project

the code with processing steps can be found here: https://github.com/ertan-somundzhu/sad-model

download the processed version with:

from huggingface_hub import snapshot_download
snapshot_download('hypersunflower/ava_speech_data_log_mel_spec', local_dir = "data/human-ava-speech", repo_type="dataset" )

load the processed data with:

import os
import numpy as np

data = np.load(os.path.abspath("data/human-ava-speech/dataset.npz"))
mel = data["mel"]
onset = data["onset"]
offset = data["offset"]
filenames = data["filenames"]