Skip to main content
Sonic provides controls for the speed, volume, and emotion of generated speech. These are available on play.cartesia.ai using the UI controls, by passing a generation_config parameter on each request, or by using SSML tags within the transcript.
generation_config is scoped to a single request — there is no persistent or account-wide setting. Include the parameter on every TTS request where you want speed, volume, or emotion guidance applied. For finer variation within a single generation (for example, shifting emotion sentence-to-sentence), use SSML tags inline in the transcript.
Sonic interprets these parameters as guidance rather than strict adjustments, to ensure natural speech. Test against your content to confirm the output matches your expectations.

​
Speed and volume controls

Guide the speed and volume of a TTS generation with the generation_config.speed and generation_config.volume parameters.
​
number
The speed of the generation, ranging from 0.6 to 1.5.
​
number
The volume of the generation, ranging from 0.5 to 2.0.
Set speed and volume via generation_config rather than SSML tags. Most well-punctuated transcripts are paced naturally without any adjustment, so treat these controls as a refinement for specific cases rather than a default.
You can also set these inside the transcript using SSML tags.

​
Emotion controls Beta

By default, the model interprets the emotional subtext in the provided transcript. Guide the emotion of a TTS generation, the way a director directs an actor, using the generation_config.emotion parameter.
Emotion tags push the model to be more emotive, but only work when the emotion is consistent with the transcript. The mismatch below is unlikely to work well:
​
string
The emotional guidance for a generation. Must be one of the values listed below — passing an emotion outside this list is not supported, and results are not guaranteed. Behavior for unrecognized values may change without notice.
The primary emotions, for which we have the most data and produce the best results, are: neutral, calm, angry, content, sad, and scared. The complete list of available emotions is: neutral, happy, excited, enthusiastic, elated, euphoric, triumphant, amazed, surprised, flirtatious, curious, content, peaceful, serene, calm, grateful, affectionate, trust, sympathetic, anticipation, mysterious, angry, mad, outraged, frustrated, agitated, threatened, disgusted, contempt, envious, sarcastic, ironic, sad, dejected, melancholic, disappointed, hurt, guilty, bored, tired, rejected, nostalgic, wistful, apologetic, hesitant, insecure, confused, resigned, anxious, panicked, alarmed, scared, proud, confident, distant, skeptical, contemplative, determined. The voices with the best emotional response are:
  • Leo (id: 0834f3df-e650-4766-a20c-5a93a43aa6e3)
  • Jace (id: 6776173b-fd72-460d-89b3-d85812ee518d)
  • Kyle (id: c961b81c-a935-4c17-bfb3-ba2239de8c2f)
  • Gavin (id: f4a3a8e4-694c-4c45-9ca0-27caf97901b5)
  • Maya (id: cbaf8084-f009-4838-a096-07ee2e6612b1)
  • Tessa (id: 6ccbfb76-1fc6-48f7-b71d-91ac6298247b)
  • Dana (id: cc00e582-ed66-4004-8336-0175b85c85f6)
  • Marian (id: 26403c37-80c1-4a1a-8692-540551ca2ae5)
View the full list of emotive voices in our Voice Library. You can also use SSML tags for emotions:

​
Nonverbalisms

Insert [laughter] in your transcript to make the model laugh.