Scholarlink

Gender recognition and age estimation based on voice play an important role in a wide range of applications, including but not limited to security systems, human-computer interaction, telecommunication services, and forensic voice analysis. We applied a light-weight Bi-LSTM neural network and combined MFCC, F0 and the first formant as the input to identify the gender and age of a speaker, the network has about 16000 (64 KB) trainable parameters. The results show that it could achieve a 98.87% accuracy in gender recognition and a 4.90 MAE in age estimation based on the TIMIT corpus.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Voice-Based Gender and Age Recognition System on Devices with Low Computational Resources

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Haoxuan Li,
  • Yonghui Rao

摘要

Gender recognition and age estimation based on voice play an important role in a wide range of applications, including but not limited to security systems, human-computer interaction, telecommunication services, and forensic voice analysis. We applied a light-weight Bi-LSTM neural network and combined MFCC, F0 and the first formant as the input to identify the gender and age of a speaker, the network has about 16000 (64 KB) trainable parameters. The results show that it could achieve a 98.87% accuracy in gender recognition and a 4.90 MAE in age estimation based on the TIMIT corpus.