mandada/forced-alignment-server

AssemblyAI Transcription Server

This Node.js/Express server provides an API to transcribe audio files using AssemblyAI's asynchronous transcription service. It includes a caching layer using Cloudflare R2 to avoid re-transcribing the same audio files.

Features

• —Asynchronous Transcription: Submits audio URLs for transcription and returns a job ID immediately.
• —Status Polling: Allows clients to poll for the transcription status.
• —Cloudflare R2 Caching: Caches completed transcriptions to reduce costs and improve response times.
• —Secure: Uses environment variables to manage API keys and other secrets.

Setup

1. 1.Install Dependencies:

    npm install

1. 1.Create `.env` file: Create a .env file in the root of the project and add the following environment variables:

    ASSEMBLYAI_API_KEY=your_assemblyai_api_key
    R2_ACCOUNT_ID=your_r2_account_id
    R2_ACCESS_KEY_ID=your_r2_access_key_id
    R2_SECRET_ACCESS_KEY=your_r2_secret_access_key
    R2_BUCKET_NAME=saintshubapp-transcribed-sermons

1. 1.Run the server:

    npm run dev

The server will start on http://localhost:3000.

Testing Endpoints

You can use curl to test the server endpoints.

1. Submit a transcription job

This will submit an audio URL to the /transcribe endpoint. The server will return a JSON object with the status and the transcription job ID.

curl -X POST -H "Content-Type: application/json" -d '{"audioUrl": "https://assemblyai-realtimestatic-prod.s3.us-west-2.amazonaws.com/static/demos/media/20130514-105844-320k.mp3"}' http://localhost:3000/transcribe

2. Extract Text from a PDF

This will submit a PDF URL to the /extract-pdf endpoint. The server will download the PDF, extract its text content, cache the result in R2, and return the full text.

curl -X POST -H "Content-Type: application/json" -d '{"pdfUrl": "https://www.sounddoctrine.net/sermons/pdf/101214.pdf"}' http://localhost:3000/extract-pdf

Expected Response:

{
  "status": "processing",
  "id": "some-transcription-id"
}

2. Check the transcription status

Use the id from the previous step to poll the /status/:id endpoint. Replace YOUR_TRANSCRIPTION_ID with the actual ID.

curl http://localhost:3000/status/YOUR_TRANSCRIPTION_ID

Expected Response (once completed):

The server will return the full transcription object from AssemblyAI, including the text and word-level timestamps.

{
  "id": "YOUR_TRANSCRIPTION_ID",
  "status": "completed",
  "text": "This is the transcribed text...",
  "words": [...]
}