Proposal of speaker change detection system considering speaker overlap = 화자 겹침을 고려한 화자 전환 검출 시스템 제안

다국어 초록 (Multilingual Abstract)

Speaker Change Detection (SCD) refers to finding the moment when the main speaker changes from one person to the next in a speech conversation. In speaker change detection, difficulties arise due to overlapping speakers, inaccuracy in the information labeling, and data imbalance. To solve these problems, TIMIT corpus widely used in speech recognition have been concatenated artificially to obtain a sufficient amount of training data, and the detection of changing speaker has performed after identifying overlapping speakers. In this paper, we propose an speaker change detection system that considers the speaker overlapping. We evaluated and verified the performance using various approaches. As a result, a detection system similar to the X-Vector structure was proposed to remove the speaker overlapping region, while the Bi-LSTM method was selected to model the speaker change system. The experimental results show a relative performance improvement of 4.6 % and 13.8 % respectively, compared to the baseline system. Additionally, we determined that a robust speaker change detection system can be built by conducting related studies based on the experimental results, taking into consideration text and speaker information.

참고문헌 (Reference)

1 R. Yin, "peaker change detection in broadcast tv using bidirectional long short term memory networks" 3827-3831, 2017

2 "WebRTC Homepage"

3 S. C. Levinson, "Turn-taking in human communication - Origins and implications for language processing" 20 : 6-14, 2016

4 H. Bredin, "TristouNet: Triplet loss for speaker turn embedding" 5430-5434, 2017

5 V. Zue, "Speech database development at MIT: TIMIT and beyond" 9 : 351-356, 1990

6 Z. Ge, "Speaker change detection using features through a neural network speaker classier" 1111-1116, 2017

7 L. Bullock, "Overlap aware diarization: Resegmentation using neural end to-end overlapped speech detection" 7114-7118, 2020

8 N. Sajjan, "Leveraging lstm models for overlap detection in multi party meetings" 5249-5253, 2018

9 H. Kim, "Framework switching of speaker overlap de tection system" 17 : 101-113, 2021

10 M. Kunesova, "Detection of overlapping speech for the purposes of speaker diarization" 247-257, 2019

11 V. Andrei, "Detecting over-lapped speech on short time frames using deep learning" 1198-1202, 2017

12 J. Park, "Data augmentation and d-vector representation methods for speaker change detection" 67-71, 2020

13 E. Kazimirova, "Automatic detection of multi speaker fragments with high time resolution" 1338-1392, 2018

14 A. G. Adam, "A new speaker change detection method for two-speaker segmentation" 3908-3911, 2002

15 D. Snyder, "2X-vectors: Robust DNN embeddings for speaker recognition" 5329-5333, 2018