Building a good speech recognition system usually requires a lot of pairing data, which poses a big challenge for low-resource languages, such as Kazakh. In recent years, unsupervised pre-training has achieved good performance in low-resource speech recognition, but it is rarely used in Kazakh and other Central and West Asian languages. In this paper, wav2vec2.0 is improved by integrating a Factorized TDNN layer to better preserve the relationship between the voice and the time step before and after the quantization, which is called wav2vec-F. The unsupervised pre-training strategy was used to learn the potential speech representation from a large number of unlabeled audio data and was applied to the cross-language ASR task, which was optimized using the noise contrast binary classification task. At the same time, speech synthesis is used to promote the performance of speech recognition. The experiment shows that wav2vec-F can effectively utilize the unlabeled data from non-target languages, and the multi-language pre-training is obviously better than the single-language pre-training. The data enhancement method using speech synthesis can bring huge benefits. Compared with the baseline model, Librispeech’s test-clean dataset has an average reduction of 1.9% in the word error rate. On the Kazakh KSC test set, the pre-training using only Kazakh reduced the word error rate by 3.8%. The pre-training of a small amount of Kazakh speech data synthesized by multi-language combined with TTS achieved a word error rate of 8.6% on the KSC test set when the labeled data were only 10 h, which was comparable to the results of the previous end-to-end model when the labeled data were 30 times less.

Unlike the traditional model, the end-to-end (E2E) ASR model does not require speech information such as a pronunciation dictionary, and its system is built through a single neural network and obtains performance comparable to that of traditional methods. However, the model requires massive amounts of training data. Recently, hybrid CTC/attention ASR systems have become more popular and have achieved good performance even under low-resource conditions, but they are rarely used in Central Asian languages such as Turkish and Uzbek. We extend the dataset by adding noise to the original audio and using speed perturbation. To develop the performance of an E2E agglutinative language speech recognition system, we propose a new feature extractor, MSPC, which uses different sizes of convolution kernels to extract and fuse features of different scales. The experimental results show that this structure is superior to VGGnet. In addition to this, the attention module is improved. By using the CTC objective function in training and the BERT model to initialize the language model in the decoding stage, the proposed method accelerates the convergence of the model and improves the accuracy of speech recognition. Compared with the baseline model, the character error rate (CER) and word error rate (WER) on the LibriSpeech test-other dataset increases by 2.42% and 2.96%, respectively. We apply the model structure to the Common Voice—Turkish (35 h) and Uzbek (78 h) datasets, and the WER is reduced by 7.07% and 7.08%, respectively. The results show that our method is close to the advanced E2E systems.

Consonant and vowel reduction are often encountered in Uyghur speech, which might cause performance degradation in Uyghur automatic speech recognition (ASR). Our recently proposed learning strategy based on masking, Phone Masking Training (PMT), alleviates the impact of such phenomenon in Uyghur ASR. Although PMT achieves remarkably improvements, there still exists room for further gains due to the granularity mismatch between masking unit of PMT (phoneme) and modeling unit (word-piece). To boost the performance of PMT, we propose multi-modeling unit training (MMUT) architecture fusion with PMT (PM-MMUT). The idea of MMUT framework is to split the Encoder into two parts including acoustic feature sequences to phoneme-level representation (AF-to-PLR) and phoneme-level representation to word-piece-level representation (PLR-to-WPLR). It allows AF-to-PLR to be optimized by an intermediate phoneme-based CTC loss to learn the rich phoneme-level context information brought by PMT. Experimental results on Uyghur ASR show that the proposed approaches improve significantly, outperforming the pure PMT (reduction WER from 24.0 to 23.7 on Read-Test and from 38.4 to 36.8 on Oral-Test respectively). We also conduct experiments on the 960-hour Librispeech benchmark using ESPnet1, which achieves about 10% relative WER on all the test sets without LM fusion comparing with the latest official ESPnet1 pre-trained model.

No abstract

In recent years, the end-to-end speech recognition model has emerged as a popular alternative to the traditional Deep Neural Network—Hidden Markov Model (DNN-HMM). This approach maps acoustic features directly onto text sequences via a single network architecture, significantly streamlining the model construction process. However, the training of end-to-end speech recognition models typically necessitates a significant quantity of supervised data to achieve good performance, which poses a challenge in low-resource conditions. The use of unsupervised representation significantly reduces this necessity. Recent research has focused on end-to-end techniques employing joint Connectionist Temporal Classification (CTC) and attention mechanisms, with some also concentrating on unsupervised presentation learning. This paper proposes a joint supervised and unsupervised multi-task learning model (JSUM). Our approach leverages the unsupervised pre-trained wav2vec 2.0 model as a shared encoder that integrates the joint CTC-Attention network and the generative adversarial network into a unified end-to-end architecture. Our method provides a new low-resource language speech recognition solution that optimally utilizes supervised and unsupervised datasets by combining CTC, attention, and generative adversarial losses. Furthermore, our proposed approach is suitable for both monolingual and cross-lingual scenarios.