Speakers Identification Using Diarization Techniques

Abstract

Research work analyses speaker voice identification and voice separation development methodologies and show an overview of the findings. Several speech recognition methods, such as Mel Frequency Cepstrum Coefficients (MFCC), Vector Quantization (VQ), Hidden Markov Model (HMM), Long Short-Term Memory (LSTM), End-to-End Neural Diarization (EEND), Generative Adversarial Networks (GANs), Convolutional Neural Networks, and Audio Embeddiment, can be used for adaptive processing with multiple speakers identification in audio data. Additionally, we addressed the uses of speaker diarization, the potential for future development, and the databases used to evaluate diarization systems.The speaker diarization method consists of seven steps, including input, front-end processing, speech activity detection, segmentation, speaker embedding, clustering post-processing, and output.Speaker identification recognizes speakers during an audio conversion, a kind of speech recognition. Diarization of the speaker is a way of recognizing the speaker in a multi-speaker audio file. And The procedure of identifying who talks when in an audio recording is known as speaker diarization. The audio file includes information from conferences, broadcast news, and any other public gathering with many speakers.