Noisy speech database for training speech enhancement algorithms and TTS models

By Cassia Valentini-Botinhao. 2017

241 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Speech & Audio, Signal Processing

How researchers see this paper

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

In Artificial Intelligence

In Speech & Audio

In Signal Processing

Citations in the corpus

Audiodec: An Open-Source Streaming High-Fidelity Neural Audio Codec (2023, methodology): “The Valentini dataset [35], which is derived from the VCTK corpus [36], is adopted for the evaluations.”
Generative Speech Coding with Predictive Variance Regularization (2021, methodology): “The evaluation dataset is composed of 30 samples from the Noisy VCTK dataset [30]: 15 clean and 15 augmented with additive noise at various SNRs (2.5, 7.5 and 12.5 dB).”
Enhancing into the Codec: Noise Robust Speech Coding with Vector-Quantized Autoencoders (2021, methodology): “To evaluate, we use the test mixtures from the VCTK dataset [20].”
Real Time Speech Enhancement in the Waveform Domain (2020, methodology): “Training We train the DEMUCS model for 400 epochs on the Valentini [18] dataset and 250 on the DNS [19] dataset.”
High-quality Speech Coding with Sample RNN (2018, methodology): “We constructed a new dataset based on roughly equal contributions from the WSJ0 set and VCTK set (clean speech from [18]), and also TSP set [30](1% of the full data set).”
Speaker-independent raw waveform model for glottal excitation (2018, methodology): “We use a multi-speaker database originally released for speech enhancement research [20], and only take the clean speech subset for these experiments.”
SpeechBrain: A General-Purpose Speech Toolkit (2021, methodology): “In Table 7 we compare the best enhancement systems available in SpeechBrain against the SotA DEMUCS model [95] on the Voicebank-DEMAND corpus [96].”
Universal Speech Enhancement with Score-based Diffusion (2022, methodology): “The only exception is with the COVL metric on the Voicebank-DEMAND data set, where nonetheless UNIVERSE is still competitive among the top-performing systems.”
High Fidelity Speech Regeneration with Application to Speech Enhancement (2021, background): “The second dataset is a standard benchmark in the speech enhancement literature [30].”
Generative adversarial network-based approach to signal reconstruction from magnitude spectrogram (2018, methodology): “1) Dataset: We used clean speech signals excerpted from [13] as the experimental data.”
Towards Achieving Robust Universal Neural Vocoding (2018, background): “background noise [30], reverberation [31], or both [32]), iii) unseen languages (Ahmaric, Swahili and Wolof) recorded in sub-optimal recording situations [33] (i.”
Discretization and Re-synthesis: an alternative method to solve the Cocktail Party Problem (2021, methodology): “Next, we carried out the speech enhancement experiment on the VCTK noisy corpus [15], trained with HuBERT discretizer sing the training set (tr 26spk).”