# Incremental Transfer Learning in Two-pass Information Bottleneck based Speaker Diarization System for Meetings

## TL;DR

This paper introduces an incremental transfer learning method to enhance the real-time factor of a two-pass information bottleneck speaker diarization system, enabling faster adaptation to new conversations with minimal performance loss.

## Contribution

The paper proposes a novel incremental transfer learning approach that updates neural network parameters using previous conversation data, significantly reducing training time in speaker diarization.

## Key findings

- Achieved 33.07% and 24.45% RTF reduction on NIST and AMI datasets.
- Maintained comparable speaker diarization performance with minor degradation.
- Demonstrated effectiveness on standard conversational meeting datasets.

## Abstract

The two-pass information bottleneck (TPIB) based speaker diarization system operates independently on different conversational recordings. TPIB system does not consider previously learned speaker discriminative information while diarizing new conversations. Hence, the real time factor (RTF) of TPIB system is high owing to the training time required for the artificial neural network (ANN). This paper attempts to improve the RTF of the TPIB system using an incremental transfer learning approach where the parameters learned by the ANN from other conversations are updated using current conversation rather than learning parameters from scratch. This reduces the RTF significantly. The effectiveness of the proposed approach compared to the baseline IB and the TPIB systems is demonstrated on standard NIST and AMI conversational meeting datasets. With a minor degradation in performance, the proposed system shows a significant improvement of 33.07% and 24.45% in RTF with respect to TPIB system on the NIST RT-04Eval and AMI-1 datasets, respectively.