34 datasets found

Speech enhancement deep learning systems usually require large amounts of training data to operate in broad conditions or real applications. This makes the adaptability of those...

Generative models have gained more and more attention in recent years for their remarkable success in tasks that required estimating and sampling data distribution to generate...

Training and test datasets were generated by contaminating the clean data with reverberation and noise.

The dataset used in the paper is a selection of 14 noisy mixtures created manually from the Voice Bank speech corpus.

The dataset used in the paper is a selection of ten British English speakers – both male and female – from the Voice Bank speech corpus, each of which has around 400 clean...

The MultiCorpus dataset is a diverse and acoustically adverse dataset generated by mixing reverberant speech and reverberant noise using speech utterances, noise segments and...

The VoiceBank+DEMAND (VBDMD) 28-speaker dataset is a publicly available and popular benchmark dataset for single-channel speech enhancement.

The VCTK corpus is an English multi-speaker dataset, with 44 hours of audio spoken by 109 native English speakers.

Using power level difference for near field dual-microphone speech enhancement

The authors created a custom dataset for their experiment, consisting of 33,000 images of 320 possible object-image combinations, with 10 possible shapes, 8 possible colors, 2...

The VoiceBank-DEMAND dataset is a standard benchmark for speech denoising systems. It consists of 28 speakers with 4 signal-to-noise ratios (SNR) (15, 10, 5, and 0 dB) and...

Diffusion-based speech enhancement with a weighted generative-supervised learning loss

The DNS challenge provides a blind test set for both non-personalized and personalized DNS models.

The DNS challenge provides clean speech, noise, impulse responses, and a training data synthesizer for both non-personalized and personalized DNS models.