Deep Speaker Embeddings for Far-Field Speaker Recognition on Short Utterances

Gusev, Aleksei; Volokhov, Vladimir; Andzhukaev, Tseren; Novoselov, Sergey; Lavrentyeva, Galina; Volkova, Marina; Gazizullina, Alice; Shulipa, Andrey; Gorlanov, Artem; Avdeeva, Anastasia; Ivanov, Artem; Kozlov, Alexander; Pekhovsky, Timur; Matveev, Yuri

doi:10.48550/arxiv.2002.06033

Cited by 7 publications

(8 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As proved in [25], in conventional training setting, mismatch of length between train and test speech can degrade performance for short utterance. In other words, as shown in [24], model trained with short segment performs better with short test utterance than model trained with relatively long speech, but with relatively poor performance with long speech. This tradeoff is very fatal to realistic settings, leading to not discriminative embedding of enrollment or test utterance.…”

Section: Meta-learning For Imbalance Length Pairmentioning

confidence: 82%

“…However, TDV shows good performance only in short length and relatively low performance in long utterance. In addition to these methods, there have been many attempts to solve this problem such as knowledge distillation [21], generative adversarial networks [22], angular margin-based method [23,24].…”

Section: Related Workmentioning

confidence: 99%

“…For the comparison with other previous state-of-the-arts models [2,3,24], we trained the model with VoxCeleb2 dataset and tested on VoxCeleb1 full dataset. We use same trial as described in [24]. For every speaker, trial is randomly generated for 100 positive pairs and 100 negative pairs.…”

Section: Speaker Verification For Short Utterancementioning

confidence: 99%

“…Furthermore, our model uses only 40-dimensional feature but the other baselines use more than twice the dimensions. As proven in [24], it means that the performance gap can be bigger if we use higher dimensional inputs. For the comparison with [2], since UtterIdNet is not publicly available, we compared it using TDV instead.…”

Section: Speaker Verification For Short Utterancementioning

confidence: 99%

See 3 more Smart Citations

Meta-Learning for Short Utterance Speaker Recognition with Imbalance Length Pairs

et al. 2020

View full text Add to dashboard Cite

In realistic settings, a speaker recognition system needs to identify a speaker given a short utterance, while the utterance used to enroll may be relatively long. However, existing speaker recognition models perform poorly with such short utterances. To solve this problem, we introduce a meta-learning scheme with imbalance length pairs. Specifically, we use a prototypical network and train it with a support set of long utterances and a query set of short utterances. However, since optimizing for only the classes in the given episode is not sufficient to learn discriminative embeddings for other classes in the entire dataset, we additionally classify both support set and query set against the entire classes in the training set to learn a well-discriminated embedding space. By combining these two learning schemes, our model outperforms existing state-of-the-art speaker verification models learned in a standard supervised learning framework on short utterance (1-2 seconds) on VoxCeleb dataset. We also validate our proposed model for unseen speaker identification, on which it also achieves significant gain over existing approaches.

show abstract

Section: Meta-learning For Imbalance Length Pairmentioning

confidence: 82%

Section: Related Workmentioning

confidence: 99%

Section: Speaker Verification For Short Utterancementioning

confidence: 99%

Section: Speaker Verification For Short Utterancementioning

confidence: 99%

See 2 more Smart Citations

Meta-Learning for Short Utterance Speaker Recognition with Imbalance Length Pairs

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Recently, deep neural network (DNN) and convolutional neural network (CNN) based speaker embedding systems have been applied to solve this problem and obtain effective performance improvements [9,10,11,12]. In [10], a raw waveform CNN-LSTM architecture was proposed to extract phoneticlevel features which can help compensate for missing phonetic information.…”

Section: Introductionmentioning

confidence: 99%

Open-Set Short Utterance Forensic Speaker Verification Using Teacher-Student Network with Explicit Inductive Bias

2020

View full text Add to dashboard Cite

In forensic applications, it is very common that only small naturalistic datasets consisting of short utterances in complex or unknown acoustic environments are available. In this study, we propose a pipeline solution to improve speaker verification on a small actual forensic field dataset. By leveraging large-scale out-of-domain datasets, a knowledge distillation based objective function is proposed for teacher-student learning, which is applied for short utterance forensic speaker verification. The objective function collectively considers speaker classification loss, Kullback-Leibler divergence, and similarity of embeddings. In order to advance the trained deep speaker embedding network to be robust for a small target dataset, we introduce a novel strategy to fine-tune the pre-trained student model towards a forensic target domain by utilizing the model as a finetuning start point and a reference in regularization. The proposed approaches are evaluated on the 1 st 48-UTD forensic corpus, a newly established naturalistic dataset of actual homicide investigations consisting of short utterances recorded in uncontrolled conditions. We show that the proposed objective function can efficiently improve the performance of teacher-student learning on short utterances and that our fine-tuning strategy outperforms the commonly used weight decay method by providing an explicit inductive bias towards the pre-trained model.

show abstract

Exploring the potential of Wav2vec 2.0 for speech emotion recognition using classifier combination and attention-based feature fusion

Nasersharif,

Namvarpour

2024

J Supercomput

View full text Add to dashboard Cite

Deep Speaker Embeddings for Far-Field Speaker Recognition on Short Utterances

Cited by 7 publications

References 31 publications

Meta-Learning for Short Utterance Speaker Recognition with Imbalance Length Pairs

Meta-Learning for Short Utterance Speaker Recognition with Imbalance Length Pairs

Open-Set Short Utterance Forensic Speaker Verification Using Teacher-Student Network with Explicit Inductive Bias

Exploring the potential of Wav2vec 2.0 for speech emotion recognition using classifier combination and attention-based feature fusion

Contact Info

Product

Resources

About