The service is running in test mode — beta testing is underway. Errors are possible.
Home/Blog/Fine-Tuning Speech in the API: Pauses, Stress, Emotions
For developers

Fine-Tuning Speech in the API: Pauses, Stress, Emotions

H
The HancVoice TeamMay 12, 2026 · 8 min read
Listen to the articleVoiced by HancVoice
00:0000:00
Fine-Tuning Speech in the API: Pauses, Stress, Emotions

HancVoice has no SSML markup — expressiveness is controlled by clear parameters of the /api/v1/tts request and by the editor's tools.

Core parameters

  • instruct — emotion and manner (in English, e.g. "calm, friendly").
  • emotion — a recorded emotion of a studio voice (joyful, sad, serious…).
  • speed and gain_db — pace and volume.
  • eq — timbre by frequency band; effect — a sound preset (podcast, audiobook…).

Pauses and stress

Pauses and emphasis are placed in the editor (the "How to read" panel) — they are saved as segments and passed to synthesis element by element. Stress is set reliably by marking the stressed vowel in a word.

A pause in the right place changes the meaning of a phrase more than the choice of voice.

Try it yourself — a free starter character pack when you sign up.

H
The HancVoice Team

The content is produced by the HancVoice team — the people who build the engine, the studio, and the marketplace.