Towards Lower Error Rates in Phoneme Recognition

141 citations recorded by the source corpus.

How researchers see this paper

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

Improving Transformer-based Speech Recognition Using Unsupervised Pre-training (2019, background): “Note downsampling would usually hurt the accuracy of speech recognition system [22, 23] and results for AISHELL-1 is usually reported on the original 16kHz data.”
A General Survey on Attention Mechanisms in Deep Learning (2022, methodology): “In the ﬁeld of audio processing, the Word Error Rate [110] and Phoneme Error Rate [111] are generally employed.”
Interpretation of Multiparty Meetings the AMI and Amida Projects (2008, background): “These so-called left context – right context (LCRC) features [13] are derived from multiple stages of MLPs that try to estimate phoneme state posterior probabilities.”
Probabilistic and Bottle-Neck Features for LVCSR of Meetings (2007, methodology): “The use of phoneme model states as classes is proposed in [6].”
Query-by-example spoken term detection using phonetic posteriorgram templates (2009, methodology): “The authors would like to thank Fred Richardson for his help training and running the BUT phonetic recognizer.”
Optimizing bottle-neck features for lvcsr (2008, methodology): “Since their introduction, several modifications targeting the input spectro gram [3, 4], the MLP structure [5] and MLP training targets [6] were propo sed.”
The language-independent bottleneck features (2012, methodology): “ANN Parameterization TRAPs-DCT features [12] were used as ANN input: The parameters are 15 log Mel-filterbank”