Proceedings ArticleDOI
Fast approach to speaker identification for large population using MLLR and sufficient statistics
Achintya Kumar Sarkar,Shakti P. Rath,Srinivasan Umesh +2 more
- pp 1-5
Reads0
Chats0
TLDR
A Maximum Likelihood Linear Regression (MLLR) based fast method to calculate the likelihood from the speaker model using the MLLR matrix that performs faster than GMM-UBM based system with some degradation in system accuracy.Abstract:
In speaker identification, most of the computational processing time is required to calculate the likelihood of the test utterance of the unknown speaker with respect to the speaker models in the database. When number of speakers in the database is in the order of 10,000 or more, then computational complexity becomes very high. In this paper, we propose a Maximum Likelihood Linear Regression (MLLR) based fast method to calculate the likelihood from the speaker model using the MLLR matrix. The proposed technique will help to quickly find the best N speakers during identification. After that final speaker identification task can be done within the N selected speakers using any conventional method of speaker identification. The comparative study of the proposed method is done in terms of processing time with the state-of-the-art GMM-UBM based system on NIST 2004 SRE. The proposed technique performs faster than GMM-UBM based system with some degradation in system accuracy.read more
Citations
More filters
Journal ArticleDOI
Fuzzy-Clustering-Based Decision Tree Approach for Large Population Speaker Identification
Yakun Hu,Dapeng Wu,Antonio Nucci +2 more
TL;DR: The key idea of the approach is to use a decision tree to hierarchically partition the whole population into groups of small size, and determine which speaker group at the leaf node a speaker under test belongs to, and apply MFCC+GMM to the selected speaker group for speaker identification.
Patent
Methods for creating and searching a database of speakers
TL;DR: In this paper, a method of performing a search of a database of speakers, including deriving a query utterance from the query speech sample, extracting query utterances statistics from the utterance, and performing Kernelized Locality-Sensitive Hashing (KLSH) using a kernel function, is presented.
Proceedings ArticleDOI
A fast two-level Speaker Identification method employing sparse representation and GMM-based methods
TL;DR: This paper proposes a two-step method that utilizes two different identification methods that use Nearest Neighbor method to decrease the search space and GMM-based SI methods to specify the target speaker.
Journal ArticleDOI
Speaker Identification using a Novel Prosody with Fuzzy based Hierarchical Decision Tree Approach
K. Manikandan,E. Chandra +1 more
TL;DR: The proposed speaker identification using a novel prosody with fuzzy based hierarchical decision tree approach and is used to modifying the limitations of existing traditional methods improves the performance of speaker identification in given population under noisy environments.
References
More filters
Journal ArticleDOI
Maximum likelihood from incomplete data via the EM algorithm
Journal ArticleDOI
Speaker Verification Using Adapted Gaussian Mixture Models
TL;DR: The major elements of MIT Lincoln Laboratory's Gaussian mixture model (GMM)-based speaker verification system used successfully in several NIST Speaker Recognition Evaluations (SREs) are described.
Journal ArticleDOI
Maximum likelihood linear regression for speaker adaptation of continuous density hidden Markov models
TL;DR: An important feature of the method is that arbitrary adaptation data can be used—no special enrolment sentences are needed and that as more data is used the adaptation performance improves.
The HTK book
TL;DR: The Fundamentals of HTK: General Principles of HMMs, Recognition and Viterbi Decoding, and Continuous Speech Recognition.