Although attention based end-to-end models have achieved promising performance in speech recognition, the multi-pass forward computation in beam-search increases inference time cost, which limits their practical applications. To address this issue, we propose a non-autoregressive end-to-end speech recognition system called LASO (listen attentively, and spell once). Because of the non-autoregressive property, LASO predicts a textual token in the sequence without the dependence on other tokens. Without beam-search, the one-pass propagation much reduces inference time cost of LASO. And because the model is based on the attention based feedforward structure, the computation can be implemented in parallel efficiently. We conduct experiments on publicly available Chinese dataset AISHELL-1. LASO achieves a character error rate of 6.4%, which outperforms the state-of-the-art autoregressive transformer model (6.7%). The average inference latency is 21 ms, which is 1/50 of the autoregressive transformer model.
In this paper we present an end-to-end speech recognition model with Transformer encoders that can be used in a streaming speech recognition system. Transformer computation blocks based on self-attention are used to encode both audio and label sequences independently. The activations from both audio and label encoders are combined with a feed-forward layer to compute a probability distribution over the label space for every combination of acoustic frame position and label history. This is similar to the Recurrent Neural Network Transducer (RNN-T) model, which uses RNNs for information encoding instead of Transformer encoders. The model is trained with the RNN-T loss well-suited to streaming decoding. We present results on the LibriSpeech dataset showing that limiting the left context for self-attention in the Transformer layers makes decoding computationally tractable for streaming, with only a slight degradation in accuracy. We also show that the full attention version of our model beats the-state-of-the art accuracy on the LibriSpeech benchmarks. Our results also show that we can bridge the gap between full attention and limited attention versions of our model by attending to a limited number of future frames.
A wide variety of adversarial attacks have been proposed and explored using image and audio data. These attacks are notoriously easy to generate digitally when the attacker can directly manipulate the input to a model, but are much more difficult to implement in the real-world. In this paper we present a universal, time invariant attack for general time series data such that the attack has a frequency spectrum primarily composed of the frequencies present in the original data. The universality of the attack makes it fast and easy to implement as no computation is required to add it to an input, while time invariance is useful for real-world deployment. Additionally, the frequency constraint ensures the attack can withstand filtering. We demonstrate the effectiveness of the attack in two different domains, speech recognition and unintended radiated emission, and show that the attack is robust against common transform-and-compare defense pipelines.
Error correction techniques have been used to refine the output sentences from automatic speech recognition (ASR) models and achieve a lower word error rate (WER) than original ASR outputs. Previous works usually use a sequence-to-sequence model to correct an ASR output sentence autoregressively, which causes large latency and cannot be deployed in online ASR services. A straightforward solution to reduce latency, inspired by non-autoregressive (NAR) neural machine translation, is to use an NAR sequence generation model for ASR error correction, which, however, comes at the cost of significantly increased ASR error rate. In this paper, observing distinctive error patterns and correction operations (i.e., insertion, deletion, and substitution) in ASR, we propose FastCorrect, a novel NAR error correction model based on edit alignment. In training, FastCorrect aligns each source token from an ASR output sentence to the target tokens from the corresponding ground-truth sentence based on the edit distance between the source and target sentences, and extracts the number of target tokens corresponding to each source token during edition/correction, which is then used to train a length predictor and to adjust the source tokens to match the length of the target sentence for parallel generation. In inference, the token number predicted by the length predictor is used to adjust the source tokens for target sequence generation. Experiments on the public AISHELL-1 dataset and an internal industrial-scale ASR dataset show the effectiveness of FastCorrect for ASR error correction: 1) it speeds up the inference by 6-9 times and maintains the accuracy (8-14% WER reduction) compared with the autoregressive correction model; and 2) it outperforms the accuracy of popular NAR models adopted in neural machine translation by a large margin.
Despite the remarkable progress recently made in distant speech recognition, state-of-the-art technology still suffers from a lack of robustness, especially when adverse acoustic conditions characterized by non-stationary noises and reverberation are met. A prominent limitation of current systems lies in the lack of matching and communication between the various technologies involved in the distant speech recognition process. The speech enhancement and speech recognition modules are, for instance, often trained independently. Moreover, the speech enhancement normally helps the speech recognizer, but the output of the latter is not commonly used, in turn, to improve the speech enhancement. To address both concerns, we propose a novel architecture based on a network of deep neural networks, where all the components are jointly trained and better cooperate with each other thanks to a full communication scheme between them. Experiments, conducted using different datasets, tasks and acoustic conditions, revealed that the proposed framework can overtake other competitive solutions, including recent joint training approaches.
We present an open-source speech corpus for the Kazakh language. The Kazakh speech corpus (KSC) contains around 335 hours of transcribed audio comprising over 154,000 utterances spoken by participants from different regions, age groups, and gender. 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 prepossessing procedures followed by the description of the database specifications. We also share our experience and challenges faced during database construction. To demonstrate the reliability of the database, we performed the preliminary speech recognition experiments. The experimental results imply that the quality of audio and transcripts are promising. To enable experiment reproducibility and ease the corpus usage, we also released the ESPnet recipe.
Integrating external language models (LMs) into end-to-end (E2E) models remains a challenging task for domain-adaptive speech recognition. Recently, internal language model estimation (ILME)-based LM fusion has shown significant word error rate (WER) reduction from Shallow Fusion by subtracting a weighted internal LM score from an interpolation of E2E model and external LM scores during beam search. However, on different test sets, the optimal LM interpolation weights vary over a wide range and have to be tuned extensively on well-matched validation sets. In this work, we perform LM fusion in the minimum WER (MWER) training of an E2E model to obviate the need for LM weights tuning during inference. Besides MWER training with Shallow Fusion (MWER-SF), we propose a novel MWER training with ILME (MWER-ILME) where the ILME-based fusion is conducted to generate N-best hypotheses and their posteriors. Additional gradient is induced when internal LM is engaged in MWER-ILME loss computation. During inference, LM weights pre-determined in MWER training enable robust LM integrations on test sets from different domains. Experimented with 30K-hour trained transformer transducers, MWER-ILME achieves on average 8.8% and 5.8% relative WER reductions from MWER and MWER-SF training, respectively, on 6 different test sets
The word error rate (WER) of an automatic speech recognition (ASR) system increases when a mismatch occurs between the training and the testing conditions due to the noise, etc. In this case, the acoustic information can be less reliable. This work aims to improve ASR by modeling long-term semantic relations to compensate for distorted acoustic features. We propose to perform this through rescoring of the ASR N-best hypotheses list. To achieve this, we train a deep neural network (DNN). Our DNN rescoring model is aimed at selecting hypotheses that have better semantic consistency and therefore lower WER. We investigate two types of representations as part of input features to our DNN model: static word embeddings (from word2vec) and dynamic contextual embeddings (from BERT). Acoustic and linguistic features are also included. We perform experiments on the publicly available dataset TED-LIUM mixed with real noise. The proposed rescoring approaches give significant improvement of the WER over the ASR system without rescoring models in two noisy conditions and with n-gram and RNNLM.
Recurrent neural network (RNN) language models (LMs) and Long Short Term Memory (LSTM) LMs, a variant of RNN LMs, have been shown to outperform traditional N-gram LMs on speech recognition tasks. However, these models are computationally more expensive than N-gram LMs for decoding, and thus, challenging to integrate into speech recognizers. Recent research has proposed the use of lattice-rescoring algorithms using RNNLMs and LSTMLMs as an efficient strategy to integrate these models into a speech recognition system. In this paper, we evaluate existing lattice rescoring algorithms along with new variants on a YouTube speech recognition task. Lattice rescoring using LSTMLMs reduces the word error rate (WER) for this task by 8\% relative to the WER obtained using an N-gram LM.
Form about four decades human beings have been dreaming of an intelligent machine which can master the natural speech. In its simplest form, this machine should consist of two subsystems, namely automatic speech recognition (ASR) and speech understanding (SU). The goal of ASR is to transcribe natural speech while SU is to understand the meaning of the transcription. Recognizing and understanding a spoken sentence is obviously a knowledge-intensive process, which must take into account all variable information about the speech communication process, from acoustics to semantics and pragmatics. While developing an Automatic Speech Recognition System, it is observed that some adverse conditions degrade the performance of the Speech Recognition System. In this contribution, speech enhancement system is introduced for enhancing speech signals corrupted by additive noise and improving the performance of Automatic Speech Recognizers in noisy conditions. Automatic speech recognition experiments show that replacing noisy speech signals by the corresponding enhanced speech signals leads to an improvement in the recognition accuracies. The amount of improvement varies with the type of the corrupting noise.