sarvamai

README

Sarvam AI Go SDK (Unofficial)

An idiomatic Go client for the Sarvam AI API.

Note: This is an unofficial, community-maintained SDK. It is not affiliated with or endorsed by Sarvam AI.

Features

• Chat: Conversational AI optimized for Indic contexts.
• Speech-to-Text: Support for both REST and WebSocket-based real-time transcription.
• Text-to-Speech: Natural synthesis with streaming capabilities.
• Text Processing: Translation, Transliteration, and Language Identification (LID).
• Document Intelligence: Digitization and extraction for complex documents.
• Batch Processing: Efficient handling of large audio volumes via Job APIs.
• Resilience: Automatic retries with exponential backoff and context-aware connection management.

Installation

go get github.com/Shreehari-Acharya/sarvamai-go

Quick Start

Initialization

Initialize the client using your API key from the Sarvam dashboard.

import "github.com/Shreehari-Acharya/sarvamai-go"

client, err := sarvamai.NewClient(sarvamai.Config{
    APIKey: "your-api-key",
})

Chat Completion

import "github.com/Shreehari-Acharya/sarvamai-go/chat"

resp, err := client.Chat.Completions(ctx, chat.ModelSarvamM, []chat.ChatMessage{
    chat.SystemMessage("You are a helpful assistant."),
    chat.UserMessage("What is the capital of India?"),
})

Speech-to-Text

import "github.com/Shreehari-Acharya/sarvamai-go/stt"

file, _ := os.Open("audio.wav")
defer file.Close()

resp, err := client.SpeechToText.Transcribe(ctx, file,
    stt.WithModel(stt.ModelSaaras),
    stt.WithLanguage(stt.LanguageHiIN),
)

Text-to-Speech Streaming

import "github.com/Shreehari-Acharya/sarvamai-go/tts"

stream, err := client.TextToSpeech.StreamConvert(ctx, tts.LanguageHiIN,
    tts.WithStreamSpeaker(tts.SpeakerShubh),
)
defer stream.Close()

stream.SendText("नमस्ते, आप कैसे हैं?")
stream.Flush()

for stream.Next() {
    chunk := stream.Current()
    // chunk.Audio contains base64 encoded data
}

Resilience and Reliability

• Retries: The SDK automatically retries on 429 Too Many Requests and 5xx errors.
• Timeout Management: REST requests use a default 30-second timeout, configurable via the client config.
• Resource Cleanup: WebSocket streams respect context cancellation to ensure underlying connections and goroutines are properly terminated.

Documentation And Examples

Full documentation and detailed examples are available in the examples/ directory.

Contributing

Contributions are welcome. Please refer to CONTRIBUTING.md for guidelines.

License

This project is licensed under the MIT License. See the LICENSE file for details.

Documentation

Overview

Package sarvamai provides a Go client for the Sarvam AI API.

Usage

Create a client with your API key:

client, err := sarvamai.NewClient(sarvamai.Config{
    APIKey: "your-api-key",
})

API Key

You can obtain an API key from the Sarvam dashboard: https://dashboard.sarvam.ai/key-management

Index

• type Client
  • func NewClient(cfg Config) (*Client, error)
• type Config

Types

type Client

type Client struct {
	Text                     *text.TextClient
	SpeechToText             *stt.STTClient
	SpeechToTextTranslate    *translate.TranslateClient
	SpeechToTextJob          *sttjob.SttJobClient
	SpeechToTextTranslateJob *translatejob.TranslateJobClient
	TextToSpeech             *tts.TTSClient
	Chat                     *chat.ChatClient
	DocumentIntelligence     *docintel.DocIntelClient
	// contains filtered or unexported fields
}

Client provides access to Sarvam AI services.

Services

The client exposes the following services:

• Text: For translation, transliteration, and language detection
• SpeechToText: For speech-to-text transcription (REST and streaming)
• TextToSpeech: For text-to-speech conversion
• Chat: For conversational AI interactions

Example

client, err := sarvamai.NewClient(sarvamai.Config{
    APIKey: "your-api-key",
})
if err != nil {
    log.Fatal(err)
}

// Use text translation
resp, err := client.Text.Translate(ctx, translate.Request{...})

// Use speech-to-text
resp, err := client.SpeechToText.Transcribe(ctx, stt.TranscribeRequest{...})

// Use text-to-speech
resp, err := client.TextToSpeech.Convert(ctx, tts.ConvertRequest{...})

// Use chat
resp, err := client.Chat.Completions(ctx, chat.ChatRequest{...})

func NewClient

func NewClient(cfg Config) (*Client, error)

NewClient creates a new Sarvam AI client with the given configuration.

Parameters

cfg: Configuration containing API key and optional settings. APIKey is required.

Returns

A new Client instance or an error if the API key is missing.

Example

client, err := sarvamai.NewClient(sarvamai.Config{
    APIKey:     "your-api-key",
    BaseURL:    "https://api.sarvam.ai",  // optional, default: https://api.sarvam.ai
    HTTPClient: &http.Client{Timeout: 30 * time.Second}, // optional
})

type Config

type Config struct {
	APIKey     string
	BaseURL    string
	HTTPClient *http.Client
	MaxRetries int
}

Config holds the configuration for the Sarvam AI client.

Fields

• APIKey: Your Sarvam AI API key. Required. Get one at https://dashboard.sarvam.ai/key-management
• BaseURL: The base URL for the API. Defaults to "https://api.sarvam.ai" if empty.
• HTTPClient: The HTTP client to use for requests. Defaults to a 30-second timeout if nil.
• MaxRetries: Maximum number of retries for transient errors (429, 5xx). Defaults to 3 if <= 0.