a demo code for the proposed spectral and prosodic acoustic feature enhancement under noisy environment
-
Updated
Aug 10, 2023 - Python
a demo code for the proposed spectral and prosodic acoustic feature enhancement under noisy environment
Finds location mentions in speech using an LSTM with prosodic features as input
ZPE-Prosody: deterministic prosody-feature primitive (F0 / energy / duration / voiced-mask). Encoder PASS at 13.0× / 0.64% RMSE on LibriSpeech; retrieval gate (PRO-C006) FAIL at p@5 0.31 vs 0.80; transfer (PRO-C005) paused-external.
In this project, we proposed a pipeline for word level stress/emphasis prediction from the speech data using prosodic features along with the spectral features.
Add a description, image, and links to the prosodic-features topic page so that developers can more easily learn about it.
To associate your repository with the prosodic-features topic, visit your repo's landing page and select "manage topics."