Investigation of an Encoder-Decoder LSTM Model on the Enhancement of Speech Intelligibility in Noise for Hearing Impaired Listeners

Thoidis I, Vrysis L, Pastiadis K, Markou K, Papanikolaou G. Investigation of an Encoder-Decoder LSTM Model on the Enhancement of Speech Intelligibility in Noise for Hearing Impaired Listeners. In: AES Convention 146. Audio Engineering Society; 2019. Paper 10206. Available from: https://aes.org/publications/elibrary-page/?id=20339

@inproceedings{Thoidis2019_20339,
  author = {Thoidis, Iordanis and Vrysis, Lazaros and Pastiadis, Konstantinos and Markou, Konstantinos and Papanikolaou, George},
  title = {{Investigation of an Encoder-Decoder LSTM Model on the Enhancement of Speech Intelligibility in Noise for Hearing Impaired Listeners}},
  booktitle = {AES Convention 146},
  note = {Paper 10206},
  year = {2019},
  month = mar,
  publisher = {Audio Engineering Society},
  url = {https://aes.org/publications/elibrary-page/?id=20339}
}

TY  - CPAPER
TI  - Investigation of an Encoder-Decoder LSTM Model on the Enhancement of Speech Intelligibility in Noise for Hearing Impaired Listeners
AU  - Thoidis, Iordanis
AU  - Vrysis, Lazaros
AU  - Pastiadis, Konstantinos
AU  - Markou, Konstantinos
AU  - Papanikolaou, George
T2  - AES Convention 146
M1  - Paper 10206
PY  - 2019
DA  - 2019/03/06
UR  - https://aes.org/publications/elibrary-page/?id=20339
PB  - Audio Engineering Society
LA  - en
AB  - Hearing impaired (HI) listeners often struggle to follow conversations when exposed in a complex acoustic environment. This is partly due to the reduced ability in recovering the target speech Temporal Envelope (ENV) cues from Temporal Fine Structure (TFS). This study investigates the enhancement of speech intelligibility in HI listeners by processing the ENV of speech signals corrupted by real-world environmental noise. An Encoder-Decoder Long Short Term Memory (LSTM) model is exploited after perceptually motivated processing stages to compensate for the important ENV characteristics of comprehensible speech for hearing impairment. The computational model is evaluated using the Short-Time Objective Intelligibility (STOI) measure for speech intelligibility. Finally, results indicate a 6% improvement in the mean STOI measure across different SNR values.
ER  -