Age estimation in short speech utterances based on bidirectional gated-recurrent neural networks

Abstract EN

Recently, age estimates from speech have received growing interest asthey are important for many applications like custom call routing, targetedmarketing, or user-profiling.

In this work, an automatic system to estimateage in short speech utterances without depending on the text is proposed.

From each utterance frame, four groups of features are extracted and then10 statistical functionals are measured for each extracted dimension of thefeatures, to be followed by dimensionality reduction usingLinear Discriminant Analysis (LDA).

Finally, bidirectional GatedRecurrent Neural Networks (G- RNNs) are used to predict speaker age.

Experiments are conducted on the VoxCeleb1 dataset to show theperformance of the proposed system, which is the first attempt to do sofor such a system.

In gender-dependent system, the Mean olute Error(MAE) of the proposed system is 9.25 years, and 10.33 years, the RootMean Square Error (RMSE)is 13.17 and 13.26, respectively, for femaleand male speakers.

In gender_ independent system, the MAE of theproposed system is 10.96 years, and the RMSE is 15.47.

The results showthat the proposed system has a good performance on short-durationutterances, taking into consideration the high noise ratio in the Recently, age estimates from speech have received growing interest asthey are important for many applications like custom call routing, targetedmarketing, or user-profiling.

In this work, an automatic system to estimateage in short speech utterances without depending on the text is proposed.

From each utterance frame, four groups of features are extracted and then10 statistical functionals are measured for each extracted dimension of thefeatures, to be followed by dimensionality reduction usingLinear Discriminant Analysis (LDA).

Finally, bidirectional GatedRecurrent Neural Networks (G- RNNs) are used to predict speaker age.

Experiments are conducted on the VoxCeleb1 dataset to show theperformance of the proposed system, which is the first attempt to do sofor such a system.

In gender-dependent system, the Mean olute Error(MAE) of the proposed system is 9.25 years, and 10.33 years, the RootMean Square Error (RMSE)is 13.17 and 13.26, respectively, for femaleand male speakers.

In gender_ independent system, the MAE of theproposed system is 10.96 years, and the RMSE is 15.47.

The results showthat the proposed system has a good performance on short-durationutterances, taking into consideration the high noise ratio in the VoxCeleb1dataset.