Yerbolat Khassanov

Yist Y. Lin, Tao Han, Haihua Xu et al.

One of limitations in end-to-end automatic speech recognition (ASR) framework is its performance would be compromised if train-test utterance lengths are mismatched. In this paper, we propose an on-the-fly random utterance concatenation (RUC) based data augmentation method to alleviate train-test utterance length mismatch issue for short-video ASR task. Specifically, we are motivated by observations that our human-transcribed training utterances tend to be much shorter for short-video spontaneous speech (~3 seconds on average), while our test utterance generated from voice activity detection front-end is much longer (~10 seconds on average). Such a mismatch can lead to suboptimal performance. Empirically, it's observed the proposed RUC method significantly improves long utterance recognition without performance drop on short one. Overall, it achieves 5.72% word error rate reduction on average for 15 languages and improved robustness to various utterance length.

Rustem Yeshpanov, Yerbolat Khassanov, Huseyin Atakan Varol

We present the development of a dataset for Kazakh named entity recognition. The dataset was built as there is a clear need for publicly available annotated corpora in Kazakh, as well as annotation guidelines containing straightforward--but rigorous--rules and examples. The dataset annotation, based on the IOB2 scheme, was carried out on television news text by two native Kazakh speakers under the supervision of the first author. The resulting dataset contains 112,702 sentences and 136,333 annotations for 25 entity classes. State-of-the-art machine learning models to automatise Kazakh named entity recognition were also built, with the best-performing model achieving an exact match F1-score of 97.22% on the test set. The annotated dataset, guidelines, and codes used to train the models are freely available for download under the CC BY 4.0 licence from https://github.com/IS2AI/KazNERD.

Muhammadjon Musaev, Saida Mussakhojayeva, Ilyos Khujayorov et al.

We present a freely available speech corpus for the Uzbek language and report preliminary automatic speech recognition (ASR) results using both the deep neural network hidden Markov model (DNN-HMM) and end-to-end (E2E) architectures. The Uzbek speech corpus (USC) comprises 958 different speakers with a total of 105 hours of transcribed audio recordings. To the best of our knowledge, this is the first open-source Uzbek speech corpus dedicated to the ASR task. To ensure high quality, the USC has been manually checked by native speakers. We first describe the design and development procedures of the USC, and then explain the conducted ASR experiments in detail. The experimental results demonstrate promising results for the applicability of the USC for ASR. Specifically, 18.1% and 17.4% word error rates were achieved on the validation and test sets, respectively. To enable experiment reproducibility, we share the USC dataset, pre-trained models, and training recipes in our GitHub repository.

Saida Mussakhojayeva, Aigerim Janaliyeva, Almas Mirzakhmetov et al.

This paper introduces a high-quality open-source speech synthesis dataset for Kazakh, a low-resource language spoken by over 13 million people worldwide. The dataset consists of about 93 hours of transcribed audio recordings spoken by two professional speakers (female and male). It is the first publicly available large-scale dataset developed to promote Kazakh text-to-speech (TTS) applications in both academia and industry. In this paper, we share our experience by describing the dataset development procedures and faced challenges, and discuss important future directions. To demonstrate the reliability of our dataset, we built baseline end-to-end TTS models and evaluated them using the subjective mean opinion score (MOS) measure. Evaluation results show that the best TTS models trained on our dataset achieve MOS above 4 for both speakers, which makes them applicable for practical use. The dataset, training recipe, and pretrained TTS models are freely available.

Yerbolat Khassanov, Saida Mussakhojayeva, Almas Mirzakhmetov et al.

We present an open-source speech corpus for the Kazakh language. The Kazakh speech corpus (KSC) contains around 332 hours of transcribed audio comprising over 153,000 utterances spoken by participants from different regions and age groups, as well as both genders. It was carefully inspected by native Kazakh speakers to ensure high quality. The KSC is the largest publicly available database developed to advance various Kazakh speech and language processing applications. In this paper, we first describe the data collection and preprocessing procedures followed by a description of the database specifications. We also share our experience and challenges faced during the database construction, which might benefit other researchers planning to build a speech corpus for a low-resource language. To demonstrate the reliability of the database, we performed preliminary speech recognition experiments. The experimental results imply that the quality of audio and transcripts is promising (2.8% character error rate and 8.7% word error rate on the test set). To enable experiment reproducibility and ease the corpus usage, we also released an ESPnet recipe for our speech recognition models.

Yerbolat Khassanov, Zhipeng Chen, Tianfeng Chen et al.

This paper addresses challenges in integrating new languages into a pre-trained multilingual automatic speech recognition (mASR) system, particularly in scenarios where training data for existing languages is limited or unavailable. The proposed method employs a dual-pipeline with low-rank adaptation (LoRA). It maintains two data flow pipelines-one for existing languages and another for new languages. The primary pipeline follows the standard flow through the pre-trained parameters of mASR, while the secondary pipeline additionally utilizes language-specific parameters represented by LoRA and a separate output decoder module. Importantly, the proposed approach minimizes the performance degradation of existing languages and enables a language-agnostic operation mode, facilitated by a decoder selection strategy. We validate the effectiveness of the proposed method by extending the pre-trained Whisper model to 19 new languages from the FLEURS dataset

Rustem Yeshpanov, Saida Mussakhojayeva, Yerbolat Khassanov

This work aims to build a multilingual text-to-speech (TTS) synthesis system for ten lower-resourced Turkic languages: Azerbaijani, Bashkir, Kazakh, Kyrgyz, Sakha, Tatar, Turkish, Turkmen, Uyghur, and Uzbek. We specifically target the zero-shot learning scenario, where a TTS model trained using the data of one language is applied to synthesise speech for other, unseen languages. An end-to-end TTS system based on the Tacotron 2 architecture was trained using only the available data of the Kazakh language. To generate speech for the other Turkic languages, we first mapped the letters of the Turkic alphabets onto the symbols of the International Phonetic Alphabet (IPA), which were then converted to the Kazakh alphabet letters. To demonstrate the feasibility of the proposed approach, we evaluated the multilingual Turkic TTS model subjectively and obtained promising results. To enable replication of the experiments, we make our code and dataset publicly available in our GitHub repository.

Saida Mussakhojayeva, Yerbolat Khassanov, Huseyin Atakan Varol

We present an expanded version of our previously released Kazakh text-to-speech (KazakhTTS) synthesis corpus. In the new KazakhTTS2 corpus, the overall size has increased from 93 hours to 271 hours, the number of speakers has risen from two to five (three females and two males), and the topic coverage has been diversified with the help of new sources, including a book and Wikipedia articles. This corpus is necessary for building high-quality TTS systems for Kazakh, a Central Asian agglutinative language from the Turkic family, which presents several linguistic challenges. We describe the corpus construction process and provide the details of the training and evaluation procedures for the TTS system. Our experimental results indicate that the constructed corpus is sufficient to build robust TTS models for real-world applications, with a subjective mean opinion score ranging from 3.6 to 4.2 for all the five speakers. We believe that our corpus will facilitate speech and language research for Kazakh and other Turkic languages, which are widely considered to be low-resource due to the limited availability of free linguistic data. The constructed corpus, code, and pretrained models are publicly available in our GitHub repository.

Madina Abdrakhmanova, Saniya Abushakimova, Yerbolat Khassanov et al.

In this paper, we study an approach to multimodal person verification using audio, visual, and thermal modalities. The combination of audio and visual modalities has already been shown to be effective for robust person verification. From this perspective, we investigate the impact of further increasing the number of modalities by adding thermal images. In particular, we implemented unimodal, bimodal, and trimodal verification systems using state-of-the-art deep learning architectures and compared their performance under clean and noisy conditions. We also compared two popular fusion approaches based on simple score averaging and the soft attention mechanism. The experiment conducted on the SpeakingFaces dataset demonstrates the superior performance of the trimodal verification system. Specifically, on the easy test set, the trimodal system outperforms the best unimodal and bimodal systems by over 50% and 18% relative equal error rates, respectively, under both the clean and noisy conditions. On the hard test set, the trimodal system outperforms the best unimodal and bimodal systems by over 40% and 13% relative equal error rates, respectively, under both the clean and noisy conditions. To enable reproducibility of the experiment and facilitate research into multimodal person verification, we made our code, pretrained models, and preprocessed dataset freely available in our GitHub repository.

Saida Mussakhojayeva, Yerbolat Khassanov, Huseyin Atakan Varol

We study training a single end-to-end (E2E) automatic speech recognition (ASR) model for three languages used in Kazakhstan: Kazakh, Russian, and English. We first describe the development of multilingual E2E ASR based on Transformer networks and then perform an extensive assessment on the aforementioned languages. We also compare two variants of output grapheme set construction: combined and independent. Furthermore, we evaluate the impact of LMs and data augmentation techniques on the recognition performance of the multilingual E2E ASR. In addition, we present several datasets for training and evaluation purposes. Experiment results show that the multilingual models achieve comparable performances to the monolingual baselines with a similar number of parameters. Our best monolingual and multilingual models achieved 20.9% and 20.5% average word error rates on the combined test set, respectively. To ensure the reproducibility of our experiments and results, we share our training recipes, datasets, and pre-trained models.

Madina Abdrakhmanova, Askat Kuzdeuov, Sheikh Jarju et al.

We present SpeakingFaces as a publicly-available large-scale multimodal dataset developed to support machine learning research in contexts that utilize a combination of thermal, visual, and audio data streams; examples include human-computer interaction, biometric authentication, recognition systems, domain transfer, and speech recognition. SpeakingFaces is comprised of aligned high-resolution thermal and visual spectra image streams of fully-framed faces synchronized with audio recordings of each subject speaking approximately 100 imperative phrases. Data were collected from 142 subjects, yielding over 13,000 instances of synchronized data (~3.8 TB). For technical validation, we demonstrate two baseline examples. The first baseline shows classification by gender, utilizing different combinations of the three data streams in both clean and noisy environments. The second example consists of thermal-to-visual facial image translation, as an instance of domain transfer.

Tingzhi Mao, Yerbolat Khassanov, Van Tung Pham et al.

Automatic speech recognition (ASR) for under-represented named-entity (UR-NE) is challenging due to such named-entities (NE) have insufficient instances and poor contextual coverage in the training data to learn reliable estimates and representations. In this paper, we propose approaches to enriching UR-NEs to improve speech recognition performance. Specifically, our first priority is to ensure those UR-NEs to appear in the word lattice if there is any. To this end, we make exemplar utterances for those UR-NEs according to their categories (e.g. location, person, organization, etc.), ending up with an improved language model (LM) that boosts the UR-NE occurrence in the word lattice. With more UR-NEs appearing in the lattice, we then boost the recognition performance through lattice rescoring methods. We first enrich the representations of UR-NEs in a pre-trained recurrent neural network LM (RNNLM) by borrowing the embedding representations of the rich-represented NEs (RR-NEs), yielding the lattices that statistically favor the UR-NEs. Finally, we directly boost the likelihood scores of the utterances containing UR-NEs and gain further performance improvement.

Zhiping Zeng, Van Tung Pham, Haihua Xu et al.

In this work, we study leveraging extra text data to improve low-resource end-to-end ASR under cross-lingual transfer learning setting. To this end, we extend our prior work [1], and propose a hybrid Transformer-LSTM based architecture. This architecture not only takes advantage of the highly effective encoding capacity of the Transformer network but also benefits from extra text data due to the LSTM-based independent language model network. We conduct experiments on our in-house Malay corpus which contains limited labeled data and a large amount of extra text. Results show that the proposed architecture outperforms the previous LSTM-based architecture [1] by 24.2% relative word error rate (WER) when both are trained using limited labeled data. Starting from this, we obtain further 25.4% relative WER reduction by transfer learning from another resource-rich language. Moreover, we obtain additional 13.6% relative WER reduction by boosting the LSTM decoder of the transferred model with the extra text data. Overall, our best model outperforms the vanilla Transformer ASR by 11.9% relative WER. Last but not least, the proposed hybrid architecture offers much faster inference compared to both LSTM and Transformer architectures.

Tingzhi Mao, Yerbolat Khassanov, Van Tung Pham et al.

In this paper, we present a series of complementary approaches to improve the recognition of underrepresented named entities (NE) in hybrid ASR systems without compromising overall word error rate performance. The underrepresented words correspond to rare or out-of-vocabulary (OOV) words in the training data, and thereby can't be modeled reliably. We begin with graphemic lexicon which allows to drop the necessity of phonetic models in hybrid ASR. We study it under different settings and demonstrate its effectiveness in dealing with underrepresented NEs. Next, we study the impact of neural language model (LM) with letter-based features derived to handle infrequent words. After that, we attempt to enrich representations of underrepresented NEs in pretrained neural LM by borrowing the embedding representations of rich-represented words. This let us gain significant performance improvement on underrepresented NE recognition. Finally, we boost the likelihood scores of utterances containing NEs in the word lattices rescored by neural LMs and gain further performance improvement. The combination of the aforementioned approaches improves NE recognition by up to 42% relatively.

Van Tung Pham, Haihua Xu, Yerbolat Khassanov et al.

The attention-based end-to-end (E2E) automatic speech recognition (ASR) architecture allows for joint optimization of acoustic and language models within a single network. However, in a vanilla E2E ASR architecture, the decoder sub-network (subnet), which incorporates the role of the language model (LM), is conditioned on the encoder output. This means that the acoustic encoder and the language model are entangled that doesn't allow language model to be trained separately from external text data. To address this problem, in this work, we propose a new architecture that separates the decoder subnet from the encoder output. In this way, the decoupled subnet becomes an independently trainable LM subnet, which can easily be updated using the external text data. We study two strategies for updating the new architecture. Experimental results show that, 1) the independent LM architecture benefits from external text data, achieving 9.3% and 22.8% relative character and word error rate reduction on Mandarin HKUST and English NSC datasets respectively; 2)the proposed architecture works well with external LM and can be generalized to different amount of labelled data.

Yerbolat Khassanov, Haihua Xu, Van Tung Pham et al.

The lack of code-switch training data is one of the major concerns in the development of end-to-end code-switching automatic speech recognition (ASR) models. In this work, we propose a method to train an improved end-to-end code-switching ASR using only monolingual data. Our method encourages the distributions of output token embeddings of monolingual languages to be similar, and hence, promotes the ASR model to easily code-switch between languages. Specifically, we propose to use Jensen-Shannon divergence and cosine distance based constraints. The former will enforce output embeddings of monolingual languages to possess similar distributions, while the later simply brings the centroids of two distributions to be close to each other. Experimental results demonstrate high effectiveness of the proposed method, yielding up to 4.5% absolute mixed error rate improvement on Mandarin-English code-switching ASR task.

Yerbolat Khassanov, Zhiping Zeng, Van Tung Pham et al.

The neural language models (NLM) achieve strong generalization capability by learning the dense representation of words and using them to estimate probability distribution function. However, learning the representation of rare words is a challenging problem causing the NLM to produce unreliable probability estimates. To address this problem, we propose a method to enrich representations of rare words in pre-trained NLM and consequently improve its probability estimation performance. The proposed method augments the word embedding matrices of pre-trained NLM while keeping other parameters unchanged. Specifically, our method updates the embedding vectors of rare words using embedding vectors of other semantically and syntactically similar words. To evaluate the proposed method, we enrich the rare street names in the pre-trained NLM and use it to rescore 100-best hypotheses output from the Singapore English speech recognition system. The enriched NLM reduces the word error rate by 6% relative and improves the recognition accuracy of the rare words by 16% absolute as compared to the baseline NLM.

Zhiping Zeng, Yerbolat Khassanov, Van Tung Pham et al.

Code-switching (CS) refers to a linguistic phenomenon where a speaker uses different languages in an utterance or between alternating utterances. In this work, we study end-to-end (E2E) approaches to the Mandarin-English code-switching speech recognition (CSSR) task. We first examine the effectiveness of using data augmentation and byte-pair encoding (BPE) subword units. More importantly, we propose a multitask learning recipe, where a language identification task is explicitly learned in addition to the E2E speech recognition task. Furthermore, we introduce an efficient word vocabulary expansion method for language modeling to alleviate data sparsity issues under the code-switching scenario. Experimental results on the SEAME data, a Mandarin-English CS corpus, demonstrate the effectiveness of the proposed methods.

Yerbolat Khassanov, Eng Siong Chng

In automatic speech recognition (ASR) systems, recurrent neural network language models (RNNLM) are used to rescore a word lattice or N-best hypotheses list. Due to the expensive training, the RNNLM's vocabulary set accommodates only small shortlist of most frequent words. This leads to suboptimal performance if an input speech contains many out-of-shortlist (OOS) words. An effective solution is to increase the shortlist size and retrain the entire network which is highly inefficient. Therefore, we propose an efficient method to expand the shortlist set of a pretrained RNNLM without incurring expensive retraining and using additional training data. Our method exploits the structure of RNNLM which can be decoupled into three parts: input projection layer, middle layers, and output projection layer. Specifically, our method expands the word embedding matrices in projection layers and keeps the middle layers unchanged. In this approach, the functionality of the pretrained RNNLM will be correctly maintained as long as OOS words are properly modeled in two embedding spaces. We propose to model the OOS words by borrowing linguistic knowledge from appropriate in-shortlist words. Additionally, we propose to generate the list of OOS words to expand vocabulary in unsupervised manner by automatically extracting them from ASR output.