DARPA TIMIT:: acoustic-phonetic continuous speech corpus CD-ROM, NIST speech disc 1-1.1

By John S. Garofolo, Lori Faith Lamel, W. Fisher, Jonathan G. Fiscus, D. Pallett, Nancy L. Dahlgren. 1993

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Speech & Audio

Citations in the corpus

Neural Autoregressive Distribution Estimation (2016, methodology): “We also measured the ability of RNADE to model small patches of speech spectrograms, extracted from the TIMIT dataset (Garofolo et al., 1993).”
WaveNet: A Generative Model for Raw Audio (2016, methodology): “8 PER on the test set, which is to our knowledge the best score obtained from a model trained directly on raw audio on TIMIT.”
Benchmarking Deep Reinforcement Learning for Continuous Control (2016, neutral): “Examples include MNIST (LeCun et al., 1998), Caltech101 (Fei-Fei et al., 2006), CIFAR (Krizhevsky & Hinton, 2009), ImageNet (Deng et al., 2009), PASCAL VOC (Everingham et al., 2010), BSDS500 (Martin et al., 2001), SWITCHBOARD (Godfrey et al., 1992), TIMIT (Garofolo et al., 1993), Aurora (Hirsch & Pe”
Attention-Based Models for Speech Recognition (2015, methodology): “All experiments were performed on the TIMIT corpus [19].”
LSTM: A Search Space Odyssey (2015, methodology): “We use the TIMIT Speech corpus (Garofolo et al., 1993) for framewise phone classiﬁcation.”
A Clockwork RNN (2014, methodology): “Each sequence contains an audio signal of one spoken word from the TIMIT Speech Recognition Benchmark (Garofolo et al., 1993).”
Forward and Reverse Gradient-Based Hyperparameter Optimization (2017, methodology): “Data for all experiments was obtained from the TIMIT phonetic recognition dataset (Garofolo et al., 1993).”
VoxCeleb2: Deep Speaker Recognition (2018, background): “…limitations: (i) they are ei-a ther obtained under controlled conditions (e.g., from telephone calls [29, 30] or acoustic laboratories [31, 32, 33]), (ii) they are manually annotated and hence limited in size [8], or (iii) not freely available to the speaker community [34, 33] (see [7]…”
VoxCeleb: A Large-Scale Speaker Identification Dataset (2017, methodology): “Some datasets contain artificial degradation in an attempt to mimic real world noise, such as those developed using the TIMIT dataset [19]: NTIMIT, (transmitting TIMIT recordings through a telephone handset) and CTIMIT, (passing TIMIT files through cellular telephone circuits).”
Unsupervised Learning of Disentangled and Interpretable Representations from Sequential Data (2017, methodology): “In both tasks (TIMIT and Aurora-4), the GMM-HMM system is built with Kaldi [34] using standard recipes.”
Describing Multimedia Content Using Attention-Based Encoder-Decoder Networks (2015, methodology): “2) Experimental Result: In [33], this attention-based speech recognizer was evaluated on the widely-used TIMIT corpus [61], closely following the procedure from [62].”
TERA: Self-Supervised Learning of Transformer Encoder Representation for Speech (2020, methodology): “TABLE VIII: Comparison of pre-training approaches between recent work and the proposed approach on TIMIT [52].”
Learning Latent Representations for Speech Generation and Transformation (2017, background): “The TIMIT acoustic-phonetic corpus [20, 21] contains broadband recordings of phonetically-balanced read speech.”
Composition of Deep and Spiking Neural Networks for Very Low Bit Rate Speech Coding (2016, background): “corpus [58] (the 10 sentences for speakers indexed 1–100).”
WikiGraphs: A Wikipedia Text - Knowledge Graph Paired Dataset (2021, background): “…(Lin et al., 2014; Vinyals et al., 2016), paired streams of speech and transcription data makes it possible to train speech recognition systems (Garofolo et al., 1993; Panay-otov et al., 2015) or text-to-speech synthesis models (Oord et al., 2016), and parallel corpus of text in different…”
Quaternion Convolutional Neural Networks for End-to-End Automatic Speech Recognition (2018, methodology): “• Compare and evaluate the effectiveness of the proposed QCNN to an equivalent real-valued model on the TIMIT [10] phonemes recognition task (Section 4).”
DeepMMSE: A Deep Learning Approach to MMSE-Based Noise Power Spectral Density Estimation (2020, methodology): “The clean speech recordings from the following speech corpora are included in the training set: the trainclean-100 set from the Librispeech corpus [39] (28 539 utterances), theCSTRVCTKcorpus [40] (42 015utterances), and the si∗ and sx∗ training sets from theTIMITcorpus [41] (3 696utterances).”