Vakyam AI Text-to-Speech

Overview

Use VakyamTTSService when your voice agent should speak Indian languages. The service calls Vakyam AI and returns streaming audio you can play through any Pipecat transport. Supported languages include Tamil, Hindi, Telugu, Marathi, Gujarati, Kannada, Bengali, and Indian English. Choose a preset voice or a custom voice you have cloned in Vakyam.

Source Repository

Package source, the Daily example, and issue tracker

PyPI Package

The pipecat-vakyam package on PyPI

Vakyam AI

Product overview, voices, and supported Indian languages

API Keys

Create and manage your Vakyam AI API keys

Installation

Install the community package. It is published separately from pipecat-ai:

uv add pipecat-vakyam

Prerequisites

Account and API key

Create an account at Vakyam AI
Copy an API key from API Keys. Keys look like vak_live_...

Set it in the environment:

export VAKYAM_API_KEY="vak_live_..."

Configuration

str

default:"None"

Vakyam API key. If omitted, the service reads VAKYAM_API_KEY.

VakyamTTSService.Settings

default:"None"

Voice, language, model, and speaking rate. See Settings.

str

default:"https://api.vakyam.ai"

Override this only for a non-production Vakyam endpoint.

bool

default:"False"

Permit http:// URLs that are not localhost. Leave this off in production.

int

default:"None"

Output rate in Hz (8000, 16000, 24000, or 48000). Leave unset to use the pipeline’s output sample rate.

TextAggregationMode

default:"TextAggregationMode.SENTENCE"

Defaults to sentence-sized chunks, which produces more natural spoken replies.

Settings

Pass these through VakyamTTSService.Settings(...). They can also be updated while the pipeline is running.

Parameter	Type	Default	Description
`model`	`str`	`"raaga-v1"`	Vakyam model to use.
`voice`	`str`	`"Archana"`	Preset name (for example `Archana`) or a custom voice ID starting with `vc_`.
`language`	`str`	`"ta-IN"`	Language code. See the table below.
`speed`	`float`	`1.0`	Speaking rate from `0.5` (slower) to `2.0` (faster).

Languages

Language	Code
Tamil	`ta-IN`
Hindi	`hi-IN`
Marathi	`mr-IN`
Telugu	`te-IN`
Indian English	`en-IN`
Gujarati	`gu-IN`
Bengali	`bn-IN`
Kannada	`kn-IN`

Voice catalogs and language coverage can change. See Voices and languages for the current list.

Usage

import os

from pipecat.pipeline.pipeline import Pipeline
from pipecat.pipeline.task import PipelineParams, PipelineTask
from pipecat_vakyam import VakyamTTSService

tts = VakyamTTSService(
    api_key=os.getenv("VAKYAM_API_KEY"),
    settings=VakyamTTSService.Settings(
        voice="Archana",
        language="ta-IN",
    ),
)

pipeline = Pipeline(
    [
        transport.input(),
        stt,
        context_aggregator.user(),
        llm,
        tts,
        transport.output(),
        context_aggregator.assistant(),
    ]
)
task = PipelineTask(pipeline, params=PipelineParams(allow_interruptions=True))

To speak Hindi instead of Tamil, change the language code:

tts = VakyamTTSService(
    settings=VakyamTTSService.Settings(
        voice="Archana",
        language="hi-IN",
    ),
)

Compatibility

Last tested with Pipecat v1.7.0 (pipecat-ai>=1.5,<2). See the changelog for updates.

Get Started

Pipecat Server

Client SDKs

Pipecat Flows

Pipecat Cloud

CLI

Pipecat Context Hub

Vakyam AI Text-to-Speech

Overview

Source Repository

PyPI Package

Vakyam AI

API Keys

Installation

Prerequisites

Account and API key

Configuration

Settings

Languages

Usage

Compatibility

​Overview

Source Repository

PyPI Package

Vakyam AI

API Keys

​Installation

​Prerequisites

​Account and API key

​Configuration

​Settings

​Languages

​Usage

​Compatibility

Overview

Installation

Prerequisites

Account and API key

Configuration

Settings

Languages

Usage

Compatibility