We present a novel integration of an instruction-tuned large language model (LLM) and end-to-end automatic speech recognition (ASR). Modern LLMs can perform a wide range of linguistic tasks within zero-shot learning when provided with a precise instruction or a prompt to guide the text generation process towards the desired task. We explore using this zero-shot capability of LLMs to extract linguistic information that can contribute to improving ASR performance. Specifically, we direct an LLM to correct grammatical errors in an ASR hypothesis and harness the embedded linguistic knowledge to conduct end-to-end ASR. The proposed model is built on the hybrid connectionist temporal classification (CTC) and attention architecture, where an instruction-tuned LLM (i.e., Llama2) is employed as a front-end of the decoder. An ASR hypothesis, subject to correction, is obtained from the encoder via CTC decoding, which is then fed into the LLM along with an instruction. The decoder subsequently takes as input the LLM embeddings to perform sequence generation, incorporating acoustic information from the encoder output. Experimental results and analyses demonstrate that the proposed integration yields promising performance improvements, and our approach largely benefits from LLM-based rescoring.

We propose a new incremental learning for end-to-end Automatic Speech Recognition (ASR) to extend the model's capacity on a new task while retaining the performance on previous ones. The proposed method is effective without accessing to the old dataset to address the issues of high retraining cost and unavailable old dataset. To achieve this, both attention distillation and knowledge distillation are applied to preserve the ability of the old model during the progressive learning. With an ASR model pre-trained on 12,000h Mandarin speech, we test our proposed method on 300h new scenario task and 1h new named entities task. Experiments show that our method yields 3.25% and 0.88% absolute Character Error Rate (CER) reduction on the new scenario, when compared with the pre-trained model and the full-data retraining baseline, respectively. It even yields a surprising 0.37% absolute CER reduction on the new scenario than the fine-tuning. For the new named entities task, our method significantly improves the accuracy compared with the pre-trained model, i.e. 16.95% absolute CER reduction. For both of the new task adaptions, the new models still maintain a same accuracy with the retraining baseline on the old tasks.

In this paper, we explore various approaches for semi-supervised learning in an end-to-end automatic speech recognition (ASR) framework. The first step in our approach involves training a seed model on the limited amount of labelled data. Additional unlabelled speech data is employed through a data-selection mechanism to obtain the best hypothesized output, further used to retrain the seed model. However, uncertainties of the model may not be well captured with a single hypothesis. As opposed to this technique, we apply a dropout mechanism to capture the uncertainty by obtaining multiple hypothesized text transcripts of an speech recording. We assume that the diversity of automatically generated transcripts for an utterance will implicitly increase the reliability of the model. Finally, the data-selection process is also applied on these hypothesized transcripts to reduce the uncertainty. Experiments on freely-available TEDLIUM corpus and proprietary Adobe's internal dataset show that the proposed approach significantly reduces ASR errors, compared to the baseline model.

ABSTRACT This paper investigates the impact of word-based RNN language models (RNN-LMs) on the performance of end-to-end automatic speech recognition (ASR). In our prior work, we have proposed a multilevel LM, in which character-based and word-based RNN-LMs are combined in hybrid CTC/attention-based ASR. Although this multilevel approach achieves significant error reduction in the Wall Street Journal (WSJ) task, two different LMs need to be trained and used for decoding, which increase the computational cost and memory usage. In this paper, we further propose a novel wordbased RNN-LM, which allows us to decode with only the wordbased LM, where it provides look-ahead word probabilities to predict next characters instead of the character-based LM, leading competitive accuracy with less computation compared to the multilevel LM. We demonstrate the efficacy of the word-based RNN-LMs using a larger corpus, LibriSpeech, in addition to WSJ we used in the prior work. Furthermore, we show that the proposed model achieves 5.1 %WER for WSJ Eval'92 test set when the vocabulary size is increased, which is the best WER reported for end-to-end ASR systems on this benchmark. Index Terms-- End-to-end speech recognition, language modeling, decoding, connectionist temporal classification, attention decoder 1. INTRODUCTION Automatic speech recognition (ASR) is currently a mature set of widely-deployed technologies that enable successful user interface applications such as voice search [1]. However, current systems lean heavily on the scaffolding of complicated legacy architectures that grew up around traditional techniques, including hidden Markov models (HMMs), Gaussian mixture models (GMMs), hybrid HMM/deep neural network (DNN) systems, and sequence discriminative training methods [2].