No abstract

No abstract

Data augmentation is one of the most effective ways to make end-to-end automatic speech recognition (ASR) perform close to the conventional hybrid approach, especially when dealing with low-resource tasks. Using recent advances in speech synthesis (text-to-speech, or TTS), we build our TTS system on an ASR training database and then extend the data with synthesized speech to train a recognition model. We argue that, when the training data amount is low, this approach can allow an end-to-end model to reach hybrid systems' quality. For an artificial low-resource setup, we compare the proposed augmentation with the semi-supervised learning technique. We also investigate the influence of vocoder usage on final ASR performance by comparing Griffin-Lim algorithm with our modified LPCNet. An external language model allows our approach to reach the quality of a comparable supervised setup and outperform a semi-supervised setup (both on test-clean). We establish a state-of-the-art result for end-to-end ASR trained on Lib-riSpeech train-clean-100 set with WER 4.3% on test-clean and 13.5% on test-other.

No abstract

This paper is a description of the Speech Technology Center (STC) automatic speech recognition (ASR) system for the "VOiCES from a Distance Challenge 2019". We participated in the Fixed condition of the ASR task, which means that the only training data available was an 80-hour subset of the Lib-riSpeech corpus. The main difficulty of the challenge is a mismatch between clean training data and distant noisy development/evaluation data. In order to tackle this, we applied room acoustics simulation and weighted prediction error (WPE) dereverberation. We also utilized well-known speaker adaptation using x-vector speaker embeddings, as well as novel room acoustics adaptation with R-vector room impulse response (RIR) embeddings. The system used a lattice-level combination of 6 acoustic models based on different pronunciation dictionaries and input features. N-best hypotheses were rescored with 3 neural network language models (NNLMs) trained on both words and sub-word units. NNLMs were also explored for out-of-vocabulary (OOV) words handling by means of artificial texts generation. The final system achieved Word Error Rate (WER) of 14.7% on the evaluation data, which is the best result in the challenge.

No abstract

With the rapid development of speech assistants, adapting server-intended automatic speech recognition (ASR) solutions to a direct device has become crucial. For on-device speech recognition tasks, researchers and industry prefer end-to-end ASR systems as they can be made resource-efficient while maintaining a higher quality compared to hybrid systems. However, building end-to-end models requires a significant amount of speech data. Personalization, which is mainly handling out-of-vocabulary (OOV) words, is another challenging task associated with speech assistants. In this work, we consider building an effective end-to-end ASR system in low-resource setups with a high OOV rate, embodied in Babel Turkish and Babel Georgian tasks. We propose a method of dynamic acoustic unit augmentation based on the Byte Pair Encoding with dropout (BPE-dropout) technique. The method non-deterministically tokenizes utterances to extend the token’s contexts and to regularize their distribution for the model’s recognition of unseen words. It also reduces the need for optimal subword vocabulary size search. The technique provides a steady improvement in regular and personalized (OOV-oriented) speech recognition tasks (at least 6% relative word error rate (WER) and 25% relative F-score) at no additional computational cost. Owing to the BPE-dropout use, our monolingual Turkish Conformer has achieved a competitive result with 22.2% character error rate (CER) and 38.9% WER, which is close to the best published multilingual system.

No abstract