ASR PERFORMANCE PREDICTION ON UNSEEN BROADCAST PROGRAMS USING CONVOLUTIONAL NEURAL NETWORKS

Abstract : In this paper, we address a relatively new task: prediction of ASR performance on unseen broadcast programs. We first propose an heterogenous French corpus dedicated to this task. Two prediction approaches are compared: a state-of-the-art performance prediction based on regression (engineered features) and a new strategy based on convolutional neural networks (learnt features). We particularly focus on the combination of both textual (ASR transcription) and signal inputs. While the joint use of textual and signal features did not work for the regression baseline, the combination of inputs for CNNs leads to the best WER prediction performance. We also show that our CNN prediction remarkably predicts the WER distribution on a collection of speech recordings.
Document type :
Conference papers
Liste complète des métadonnées

Cited literature [25 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-01709779
Contributor : Laurent Besacier <>
Submitted on : Thursday, February 15, 2018 - 11:43:02 AM
Last modification on : Thursday, April 4, 2019 - 10:18:05 AM
Document(s) archivé(s) le : Sunday, May 6, 2018 - 3:55:14 AM

File

20180214040820_904429_2819.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01709779, version 1

Citation

Zied Elloumi, Laurent Besacier, Olivier Galibert, Juliette Kahn, Benjamin Lecouteux. ASR PERFORMANCE PREDICTION ON UNSEEN BROADCAST PROGRAMS USING CONVOLUTIONAL NEURAL NETWORKS. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Apr 2018, Calgary, Alberta, Canada. ⟨hal-01709779⟩

Share

Metrics

Record views

519

Files downloads

267