Sarvam

Service: sarvam. A new setup starts from:

1{
2  "service": "sarvam",
3  "model": "bulbul:v3",
4  "voice": "simran",
5  "language": "en-IN"
6}

#Set it

On the flow's Voice tab, choose Speaking, then Sarvam. Or from code:

1await voxio.flows.update(flowKey, {
2  tts: {"service":"sarvam","model":"bulbul:v3","voice":"simran","language":"en-IN"},
3});

#Settings

Settings marked “under More settings” are behind More settings in the dashboard.

modelstring

Model. Which voice model speaks. Hover an option to compare. Default: "bulbul:v3".

voicestring

Voice. The voice callers hear. Default: "simran". Only applies to some models.

languagestring

Language. The language the agent speaks in. Default: "en-IN".

pacenumber (×)

Speed. How fast the agent talks. From 0.5 to 2. Default: 1.

temperaturenumber

Expressiveness (under More settings). Higher sounds livelier; lower sounds steadier. From 0.01 to 1. Default: 0.6. Only applies to some models.

pitchnumber

Pitch (under More settings). Higher or lower voice. From -1 to 1. Default: 0. Only applies to some models.

loudnessnumber

Loudness (under More settings). How loud the voice is. From 0.1 to 3. Default: 1. Only applies to some models.

enable-preprocessingboolean

Normalise text (under More settings). Reads numbers and abbreviations naturally. Default: false. Only applies to some models.

min-buffer-sizenumber (chars)

Start speaking after (under More settings). Lower starts sooner; higher sounds smoother. From 30 to 200. Default: 50.

max-chunk-lengthnumber (chars)

Chunk length (under More settings). Longest piece of text spoken at once. From 50 to 500. Default: 150.

dictionary-idstring

Pronunciation dictionary (under More settings). Your own pronunciations for names and brands. Only applies to some models.

#Choices

#Next steps

• Speaking: compare the other providers.
• Turn-taking: when the agent answers.