An Algorithm for Intelligibility Prediction of Time–Frequency Weighted Noisy Speech

By Cees H. Taal, Richard Christian Hendriks, Richard Heusdens, Jesper Højvang Jensen. 2011

2,397 citations recorded by the source corpus.

Subfields: Cognitive Neuroscience, Computational Mechanics, 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 Cognitive Neuroscience

In Computational Mechanics

In Speech & Audio

In Signal Processing

Citations in the corpus

Noise Estimation for Generative Diffusion Models (2021, methodology): “Even though our method results in a small decrease in the MCD [18], we demonstrate a large improvement in both PESQ [26] and STOI [33].”
Time-Frequency Masking in the Complex Domain for Speech Dereverberation and Denoising (2017, positive): “It is important to know that both PESQ and STOI have been shown to be highly correlated with speech quality and intelligibility of human listeners, respectively.”
Real Time Speech Enhancement in the Waveform Domain (2020, background): “We use the causal D EMUCS version and report PESQ and STOI for each of the methods.”
NaturalSpeech: End-to-End Text-to-Speech Synthesis With Human-Level Quality (2022, background): “1) Judgment Guideline: While there are some objective metrics to measure the quality gap between the generated speech and human recordings, such as PESQ [30], STOI [31], SI-SDR [32], they are not reliable to measure the perception quality in TTS [33], [34].”
Full-Capacity Unitary Recurrent Neural Networks (2016, background): “STOI is designed to correlate well with human intelligibility of speech, and takes on values between 0 and 1, with a higher score indicating higher intelligibility [20].”
Composition of Deep and Spiking Neural Networks for Very Low Bit Rate Speech Coding (2016, methodology): “We evaluated HMM, NN coding and the Speex coding using the short-time objective intelligibility (STOI) measure [70].”
End-To-End Source Separation With Adaptive Front-Ends (2017, methodology): “We compare the separation performance for each frontend transform in terms of the median BSS EVAL metrics [14] and the STOI intelligibility index [15].”
Speech Enhancement in Multiple-Noise Conditions Using Deep Neural Networks (2016, methodology): “On the other hand, we observe that DNN results in substantial gain in STOI at low SNRs. BED again outperforms all other methods where 15 − 16% improvement in STOI over noisy speech is observed at − 5 and 0 dB.”
Complex Ratio Masking for Monaural Speech Separation (2015, methodology): “The separated speech signals from each approach are evaluated with three objective metrics, namely the perceptual evaluation of speech quality (PESQ) [29], the short-time objective intelligibility (STOI) score [30], and the frequency-weighted segmental SNR (SNRfw) [31].”
ReVISE: Self-Supervised Speech Resynthesis with Visual Input for Universal and Generalized Speech Enhancement (2022, methodology): “Finally, to evaluate reconstruction of low-level details as typically done in speech denoising or source separation studies, we include ESTOI [51] and Mel cepstral distortion (MCD) [28].”
Gated Residual Networks with Dilated Convolutions for Supervised Speech Separation (2018, positive): “On the trained speakers, the proposed GRN yields around 1.6% STOI improvements compared with the LSTM on average.”
Joint Phoneme Alignment and Text-Informed Speech Separation on Highly Corrupted Speech (2020, methodology): “We also compute the Perceptual Evaluation of Speech Quality (PESQ) [26] and the Short-Time Objective Intelligibility (STOI) measure [27].”
EasyCom: An Augmented Reality Dataset to Support Algorithms for Easy Communication in Noisy Environments (2021, positive): “ESTOI has been reported to, additionally, accurately predict intelligibility when highly modulated noise is present.”
Semantic Communications: Principles and Challenges (2021, methodology): “Therefore, the metrics, such as perceptual evaluation of speech quality (PESQ) [45], short-time objective intelligibility (STOI) [46], and perceptual objective listening quality assessment (POLQA) [47], can be adopted to measure the global semantic content of speech signals, which comprehensively ev”
A New Framework for CNN-Based Speech Enhancement in the Time Domain (2019, positive): “The DNN model improves the STOI score by 16.6% at the −5 dB SNR, which is a difficult and untrained SNR condition.”
Personalized speech enhancement: new models and Comprehensive evaluation (2021, methodology): “Therefore, we rely on the word error rate (WER), deletion error rate (DEL), short-time objective intelligibility (STOI) [27] and DNSMOS [28].”