sarvamai/batch-api-typescript

sarvamai/batch-api-typescript

Language: TypeScript

Stars: 0

Forks: 1

Open issues: 1

Created: 2025-02-17T13:49:13Z

Pushed: 2025-02-19T09:41:23Z

Default branch: main

Fork: no

Archived: no

README:

Sarvam AI Speech-to-Text Batch Processing

This TypeScript script allows you to process multiple audio files using Sarvam AI's Speech-to-Text API. It handles file upload, job monitoring, and result download.

Prerequisites

1. Node.js (v14 or later) – Download from nodejs.org 2. npm (Node Package Manager) – Included with Node.js 3. A Sarvam AI API key – Obtain from dashboard.sarvam.ai 4. Audio files in WAV format

Setup

1. Clone and Install Dependencies

git clone
cd batch-api-typescript
npm install

2. Install Required Packages

npm install axios fs winston @azure/storage-blob mime-types typescript @types/node @types/mime-types @types/winston @azure/storage-file-datalake

3. Initialize TypeScript Configuration

npx tsc --init

4. Create Required Directories

mkdir src
cd src
mkdir data

The data directory stores the downloaded transcription results, and src is for TypeScript source files.

5. Create `index.ts` in the `src` Directory

Create a new file named index.ts inside the src folder and paste your provided script.

Configuration

1. API Key Setup

• Get your API key from dashboard.sarvam.ai
• Replace the API_SUBSCRIPTION_KEY value in index.ts:

const API_SUBSCRIPTION_KEY = 'API_SUBSCRIPTION_KEY';

2. Audio Files

• Update the localFiles array in main() with your audio file paths:

const localFiles = [
'/Users/adityam/Repos/batch-api-typescript/medical_domain_test (1).wav',
];

3. Running the Script

1. Compile TypeScript

npx tsc

This will generate a index.js file.

2. Run the Script

node index.js

Process Flow

1. The script initializes a new job with Sarvam AI 2. Uploads specified audio files to Azure storage 3. Starts the transcription job 4. Monitors job status until completion 5. Downloads transcription results to the data directory

Output

• Transcription results will be saved in the data directory.
• Each audio file will have a corresponding JSON file containing the transcription.
• The JSON files include:
• Transcribed text
• Confidence scores
• Timing information

Troubleshooting

1. Authentication Errors

• Verify your Sarvam AI API key is correct.
• Check if the API key has expired.

2. File Upload Errors

• Ensure audio files are in WAV format.
• Verify file paths are correct.
• Check file permissions.

3. TypeScript Errors

• Run npx tsc to check for compilation errors.
• Ensure all dependencies are properly installed.

Important Notes

• Make sure your account is whitelisted to use the API.
• Only WAV/MP3 audio files are supported.
• Monitor the console output for detailed progress information.