Skip to main content

A lightweight package for common metrics in speech processing

Project description

Speech Metrics

This package includes the common metrics used in speech processing, especially compatible with SUPERB: Speech processing Universal PERformance Benchmark. The motivation of this package is to make metric calculation disentangled from S3PRL, whose codebase and package are both too heavyweight for metric calculation. Also, some metrics in S3PRL require traditional speech processing file format (e.g. RTTM and DER) which is less pythonic and user-friendly nowadays. As a result, this package aims to extract the core logic from the existing codebases and provide a standalone python package with minimal dependencies, so that it can be used everywhere easily.

Supported Metrics

  • WER
  • CER
  • PER
  • DER
  • PESQ
  • STOI
  • SISDRi
  • SacreBLEU
  • Slot Type F1
  • Slot Value CER

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

speech_metrics-0.2.2.tar.gz (35.8 kB view hashes)

Uploaded Source

Built Distribution

speech_metrics-0.2.2-py3-none-any.whl (41.5 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page