An Initialization Scheme for Meeting Separation with Spatial Mixture Models

04/04/2022
by   Christoph Boeddeker, et al.
0

Spatial mixture model (SMM) supported acoustic beamforming has been extensively used for the separation of simultaneously active speakers. However, it has hardly been considered for the separation of meeting data, that are characterized by long recordings and only partially overlapping speech. In this contribution, we show that the fact that often only a single speaker is active can be utilized for a clever initialization of an SMM that employs time-varying class priors. In experiments on LibriCSS we show that the proposed initialization scheme achieves a significantly lower Word Error Rate (WER) on a downstream speech recognition task than a random initialization of the class probabilities by drawing from a Dirichlet distribution. With the only requirement that the number of speakers has to be known, we obtain a WER of 5.9 the estimated speaker activity from the mixture model serves as a diarization based on spatial information.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
05/02/2022

A Meeting Transcription System for an Ad-Hoc Acoustic Sensor Network

We propose a system that transcribes the conversation of a typical meeti...
research
04/08/2019

Improved Speaker-Dependent Separation for CHiME-5 Challenge

This paper summarizes several follow-up contributions for improving our ...
research
06/25/2020

Speaker-Conditional Chain Model for Speech Separation and Extraction

Speech separation has been extensively explored to tackle the cocktail p...
research
01/14/2021

Speaker activity driven neural speech extraction

Target speech extraction, which extracts the speech of a target speaker ...
research
03/24/2021

Blind Speech Separation and Dereverberation using Neural Beamforming

In this paper, we present the Blind Speech Separation and Dereverberatio...
research
05/31/2023

UNSSOR: Unsupervised Neural Speech Separation by Leveraging Over-determined Training Mixtures

In reverberant conditions with multiple concurrent speakers, each microp...
research
11/30/2018

Neural separation of observed and unobserved distributions

Separating mixed distributions is a long standing challenge for machine ...

Please sign up or login with your details

Forgot password? Click here to reset