38 datasets found

The Corpus of Spontaneous Japanese: Its design and evaluation [30] is a dataset of spontaneous Japanese speech.

The MSR dataset is a widely used vulnerability detection dataset, consisting of 10,900 vulnerable examples and 177,736 non-vulnerable examples.

The Korean GEC dataset, which includes text-level errors and is used to construct the ASR Post-Processing dataset.

The proposed text-level error type classification criteria, which considers 13 text-level errors that can occur in speech recognition situations.

The proposed speech-level error type classification criteria, which considers 24 sub-types for noise error and 13 sub-types for speaker characteristics.

The proposed Error Explainable Benchmark (EEB) dataset, which considers both speech- and text-level error types, to diagnose and validate ASR models and post-processors.

The DeepSpeech dataset used for evaluation of the proposed watermarking scheme.

The proposed black-box model watermarking framework for protecting the IP of ASR models.

Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units

Unsupervised Pre-Training for Speech Recognition

The AISHELL-1 dataset is a Mandarin speech corpus, consisting of 178 hours of speech, with 11 domains and 400 speakers from different accent areas in China.

A transformer-based Whisper Bangla ASR model

Few-shot learning evaluation of universal representations of speech

Large-scale dataset of L1 and L2 scripted and spontaneous transcripts and recordings

The dataset used in this paper is BD-4SK-ASR, an experimental dataset which is used in the first attempt in developing an ASR system for Sorani Kurdish.

The Wall Street Journal dataset is used for syntactic linearization. It contains a large corpus of news articles with their corresponding syntactic trees.

Language models have been commonly adopted to boost the performance of automatic speech recognition (ASR) particularly in domain adaptation tasks. Conventional way of LM...

The dataset used in this paper is a Conv1D equipped ASR model deployed on mobile devices.

Libriheavy is a large-scale ASR corpus consisting of 50,000 hours of read English speech derived from LibriVox. To the best of our knowledge, Libriheavy is the largest...

The ATIS dataset is a benchmark dataset for spoken language understanding, consisting of audio recordings and corresponding manual transcripts about humans asking for flight...