Deep Normalization for Speaker Vectors
نویسندگان
چکیده
Deep speaker embedding has demonstrated state-of-the-art performance in recognition tasks. However, one potential issue with this approach is that the vectors derived from deep models tend to be non-Gaussian for each individual speaker, and non-homogeneous distributions of different speakers. These irregular can seriously impact performance, especially popular PLDA scoring method, which assumes homogeneous Gaussian distribution. In article, we argue require normalization, propose a normalization based on novel discriminative flow (DNF) model. We demonstrate effectiveness proposed experiments using widely used SITW CNCeleb corpora. these experiments, DNF-based delivered substantial gains also showed strong generalization capability out-of-domain tests.
منابع مشابه
Deep Speaker Vectors for Semi Text-independent Speaker Verification
Recent research shows that deep neural networks (DNNs) can be used to extract deep speaker vectors (d-vectors) that preserve speaker characteristics and can be used in speaker verification. This new method has been tested on text-dependent speaker verification tasks, and improvement was reported when combined with the conventional i-vector method. This paper extends the d-vector approach to sem...
متن کاملSource normalization for language-independent speaker recognition using i-vectors
Source-normalization (SN) is an effective means of improving the robustness of i-vector-based speaker recognition for under-resourced and unseen cross-speech-source evaluation conditions. The technique of source-normalization estimates directions of undesired within-speaker variation more accurately than traditional methods when cross-source variation is not explicitly observed from each speake...
متن کاملImproved Speaker Markov Modelling for Unsupervised Speaker Normalization
We propose new methods of improved speech recognition with speaker-variable Information. Hidden Markov Model-based recognizers which are trained by reference speaker(s) (RS) are normalized by our two different approaches to give a better speaker-independent recognition rate. Our normalization methods are based on the same principle of inter-speaker Markov mapping. This mapping gives inter-speak...
متن کاملSpeaker independent acoustic modeling using speaker normalization
This paper proposes a novel speaker-independent (SI) modeling for spontaneous speech data from multiple speakers. The SI acoustic model parameters are estimated by individual training for inter-speaker variability and for intraspeaker phonetically related variation in order to obtain a more accurate acoustic model. The linear transformation technique is used for the speaker normalization to ext...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: IEEE/ACM transactions on audio, speech, and language processing
سال: 2021
ISSN: ['2329-9304', '2329-9290']
DOI: https://doi.org/10.1109/taslp.2020.3039573