Skip to main content

Methodology

Comparison rules for fair voice tests

vozses uses a consistent workflow so every provider is judged on the same script and playback settings.

  • Single input text per comparison session.
  • Standard MP3 output format and fixed sample rate.
  • Latency recorded for each generation.
  • Provider, voice ID, and settings are displayed.
  • Normalization is planned for v0.1 (LUFS target).