Metadata-Version: 2.4
Name: livekit-plugins-sarvam
Version: 1.5.3
Summary: Agent Framework plugin for services using Sarvam.ai's API.
Project-URL: Documentation, https://docs.livekit.io
Project-URL: Website, https://livekit.io/
Project-URL: Source, https://github.com/livekit/agents
Author-email: LiveKit <hello@livekit.io>
License-Expression: Apache-2.0
Keywords: audio,indian-languages,indic,livekit,realtime,sarvam,stt,tts,video,webrtc
Classifier: Intended Audience :: Developers
Classifier: License :: OSI Approved :: Apache Software License
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3 :: Only
Classifier: Programming Language :: Python :: 3.10
Classifier: Topic :: Multimedia :: Sound/Audio
Classifier: Topic :: Multimedia :: Video
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Requires-Python: >=3.10.0
Requires-Dist: livekit-agents[codecs,openai]>=1.5.3
Requires-Dist: numpy>=1.26
Description-Content-Type: text/markdown

# Sarvam.ai Plugin for LiveKit Agents

Support for [Sarvam.ai](https://sarvam.ai)'s Indian-language voice AI services in LiveKit Agents.

## Features

- **Speech-to-Text (STT)**: Convert audio to text using Sarvam's "Saarika" models. See the [STT docs](https://docs.livekit.io/agents/integrations/stt/sarvam/) for more information.
- **Text-to-Speech (TTS)**: Convert text to audio using Sarvam's "Bulbul" models. See the [TTS docs](https://docs.livekit.io/agents/integrations/tts/sarvam/) for more information.
- **LLM (Chat Completions)**: OpenAI-compatible chat-completions support for `sarvam-30b`, `sarvam-30b-16k`, `sarvam-105b` and `sarvam-105b-32k` including tool calling.

## Installation 

```bash
pip install livekit-plugins-sarvam
```

## Pre-requisites

You'll need an API key from Sarvam.ai. It can be set as an environment variable: `SARVAM_API_KEY`

## LLM Usage

```python
from livekit.plugins import sarvam

llm = sarvam.LLM(
    model="sarvam-30b",  # or `sarvam-30b-16k`, `sarvam-105b`, `sarvam-105b-32k`
)
```

`sarvam-30b`, `sarvam-30b-16k`, `sarvam-105b`, and `sarvam-105b-32k` support tool calling via the standard OpenAI-style `tools` and
`tool_choice` parameters.
