sarvamai-go SDK Documentation
Speech-to-Text
Edit on GitHub

Overview

Speech-to-Text APIs: REST, real-time streaming, and batch jobs.

Packages

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

Clients and methods

client.SpeechToText (stt)

client.SpeechToTextJob (sttjob)

Batch job workflow

  1. Call Initialize to create a job and set options.
  2. Call GetUploadLinks to get presigned upload URLs.
  3. Upload audio files to those URLs.
  4. Call Start to trigger processing.
  5. Poll with GetStatus until terminal state.
  6. Call GetDownloadLinks to fetch output URLs.

Model and mode rules (SDK validation)

RuleResult
ModelSaaras + Mode*Valid
ModelSaarika + Mode*Validation error
model omitted + Mode*Valid in SDK validation (uses saaras:v3 model spec for mode checks)

Language set rules (SDK validation)

Model contextAllowed language set
ModelSaarikalanguages.SaarikaLanguages
ModelSaaraslanguages.SaarasLanguages
model omittedlanguages.SaarasLanguages (validation default)

Other important combination rules

CombinationResult
WithNumSpeakers(n) + WithDiarization(true)Valid
WithNumSpeakers(n) without diarization enabledValidation error

DetectLanguage

`Text.DetectLanguage` for language and script identification.

Transcribe (REST)

`SpeechToText.Transcribe` for file-based speech recognition.

On this page

PackagesClients and methodsclient.SpeechToText (stt)client.SpeechToTextJob (sttjob)Batch job workflowModel and mode rules (SDK validation)Language set rules (SDK validation)Other important combination rules