Published March 13, 2026 | Version 1
Preprint Open

The influence of prosodic-acoustic features in automatic speech recognition of two inland Brazilian dialects: A pilot study - Supplementary materials

Authors/Creators

  • 1. ROR icon Universidade Estadual da Paraíba
  • 2. ROR icon Universidade Federal de Pernambuco
  • 3. ROR icon Universidade Estadual Paulista (Unesp)
  • 4. Companhia Pernambucana de Saneamento (COMPESA)

Description

Spplemantry materials contained in this repository:

  1. Script for realignment of the first-step forced alignment 
  2. Script for prosodic-acoustic feature extraction
  3. Spreadsheet of the acoustic features
  4. Linear Mixed-Effects Models results based on the acoustic features
  5. Report of automatic accent classification for each used machine learning algorithm

Files

CadLin_Preprint_Data_03_prosodic_features.txt

Files (202.2 kB)

Name Size Download all
md5:c2106cdc9cd26f128c3679f34d3d506c
66.3 kB Download
md5:9af391511680b7db2c14d2c8b505c0ac
84.9 kB Download
md5:9ac26a74ab94bd951a5ce9258b900bfc
36.2 kB Preview Download
md5:3602a70fff9a33a5f157dd8e32c9fdff
9.0 kB Preview Download
md5:d419ebe398dcbb69a83b01621f392900
5.8 kB Preview Download

Additional details

Funding

National Council for Scientific and Technological Development
Reconhecimento Automático de Sotaque Estrangeiro para Fins forenses 307010/2022-8