160 datasets found

The AMI Meeting Corpus was collected in three instrumented rooms with meeting conversations. Each room has two microphone arrays to collect 100 hours of far-field...

Open-domain neural dialogue generation (Vinyals and Le, 2015; Sordoni et al., 2015; Li et al., 2016a; Mou et al., 2016; Serban et al., 2016a; Asghar et al., 2016; Mei et al.,...

The Voice Bank Corpus is a large regional accent speech database containing over 10 hours of speech data from 20 speakers.

A dataset for Sanskrit ASR

A new Sanskrit speech corpus and a large-vocabulary ASR system for Sanskrit

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...

The Google Speech Command Dataset is a dataset for keyword spotting, which is a task in speech recognition. The dataset contains 12 classes, including 10 keywords and two extra...

The wav2vec 2.0 dataset is a self-supervised learning dataset for speech recognition tasks.

The CHiME-4 dataset is a large-scale speech recognition dataset, containing over 2 hours of speech from 6 channels.

Aurora-4 dataset is a broadband corpus designed for noisy speech recognition tasks based on the Wall Street Journal (WSJ0) corpus.

The Switchboard corpus is a dataset of speech recordings from a switchboard, which is a device that allows multiple people to speak at the same time.

The dataset used in the paper is the Libri-Light dataset, which is a subset of the LibriSpeech dataset. The authors used this dataset to pre-train their proposed dual-mode ASR...

The dataset used for the task of mispronunciation detection for second language learners.

The proposed method uses the acoustic unit (AU) as the intermediary feature for both mispronunciation detection and correction.

The HKUST dataset is a large dataset of speech recordings, each containing a single speaker speaking a sentence.

The WSJ dataset is a large dataset of speech recordings, each containing a single speaker speaking a sentence.

The TIMIT acoustic-phonetic continuous speech corpusCD-ROM contains a large collection of speech samples from 250 male and 250 female speakers.

SpecAugment is a data augmentation method for automatic speech recognition, which masks the mel-spectrogram along the time and frequency axes.

MixSpeech is a data augmentation method for automatic speech recognition, which trains an ASR model by taking a weighted combination of two different speech features as the...

The dataset used for training the keyword spotting model is the ESC: Dataset for Environmental Sound Classification, and the Speech Commands Dataset.