Know Your Models

Voice models

ElevenLabs, OpenAI · Lesson 13 of 14

Natural speech out, and typed-up recordings in.

Text-to-speech reads your words aloud in a natural voice.

Speech-to-text turns recordings into searchable notes.

Voice cloning needs the real person's permission, every time.

Short, clean audio gives the best results.

Best at

  • Natural narration
  • Accurate transcription
  • Cheap per minute

Watch out for

  • Consent rules around cloning
  • Struggles with heavy background noise

Try it yourself

Turn a safety handout into a two-minute audio briefing for the crew.

Where we use it in the Academy

Stephanie Teaches narration, including the Boston accent mode.

Next: Embedding models