ISCA Archive Interspeech 2022
ISCA Archive Interspeech 2022

VoiceLab: Software for Fully Reproducible Automated Voice Analysis

David Feinberg

There's a problem with acoustic analyses because you often need to hand adjust parameters meaning you can only process them individually or in small batches. This creates two key problems. First, it compromises the reproducibility of measurements because setting parameters by hand requires specialist knowledge and is often poorly documented. Second, it means that you can't easily process large samples of voices, which creates bottlenecks in workflows. This issue is compounded by researchers looking to use increasingly large and diverse samples. To address these issues, VoiceLab software offers automated acoustical analysis and automatically logs analysis parameters. VoiceLab analyses are fully reproducible and require little to no knowledge about acoustical analysis from the user. Analysis parameters can also be manually adjusted by experts. VoiceLab is used primarily by researchers studying person perception, creating reproducible voice manipulations, developing voices for conversational agents, and creating feature sets for machine learning.


doi: 10.21437/Interspeech.2022-113

Cite as: Feinberg, D. (2022) VoiceLab: Software for Fully Reproducible Automated Voice Analysis. Proc. Interspeech 2022, 351-355, doi: 10.21437/Interspeech.2022-113

@inproceedings{feinberg22_interspeech,
  author={David Feinberg},
  title={{VoiceLab: Software for Fully Reproducible Automated Voice Analysis}},
  year=2022,
  booktitle={Proc. Interspeech 2022},
  pages={351--355},
  doi={10.21437/Interspeech.2022-113}
}