项目涉及大量的跟踪和错误方法、调谐等。该模型经过良好的训练，能够区分男性和女性的声音，并且具有100%的准确度。该模型经过调整，能够以70%以上的准确率检测情绪。通过包含更多用于培训的音频文件，可以提高准确性。 项目详细说明见文档。 Datasets: Made use of two different datasets: RAVDESS. This dataset includes around 1500 audio file input from 24 different actors. 12 male and 12 female where these actors record short audios in 8 different emotions i.e 1 = neutral, 2 = calm, 3 = happy, 4 = sad, 5 = angry, 6 = fearful, 7 = disgust, 8 = surprised. Each audio file is named in such a w