/amazon-transcribe-call

Shows how to use the Amazon Transcribe API to transcribe a phone conversation recorded with the Nexmo Voice API.

Primary LanguageJavaScriptMIT LicenseMIT

Vonage Voice API Call Transcription with Amazon Transcribe

This is example code for a tutorial that shows you how to transcribe a phone call automatically using the Amazon Transcribe API. You'll need two handsets with two different phone numbers to test this.

It uses the Vonage Voice API to initiate and record the call. The call audio is created in your local recordings folder and uploaded to an S3 bucket.

Amazon Cloudwatch triggers a serverless Lambda function when the transcription job has completed. Transcripts are created in S3 and downloaded to your application's transcripts folder.

Welcome to Vonage

If you're new to Vonage, you can sign up for a Vonage account and get some free credit to get you started.

Setup

Install the Vonage CLI

Run the following at a terminal prompt to install the CLI and configure it with your api_key and api_secret, which you will find in the Developer Dashboard:

npm install -g nexmo-cli
vonage config:set --apiKey=<API_KEY> --apiSecret=<API_SECRET>

Purchase a Vonage number

If you don't already have one, buy a Vonage virtual number to receive inbound calls.

List available numbers (replace GB with your two-character country code):

vonage numbers:search GB

Purchase one of the numbers:

vonage numbers:buy 447700900001

Create a Voice API application

Use the CLI to create a Voice API application with the webhooks that will be responsible for answering a call on your Vonage number (/webhooks/answer) and logging call events (/webhooks/event), respectively. Replace example.com in the following command with your own public-facing URL host name (consider using ngrok for testing purposes, and if you do use it, run it now to get the temporary URLs that ngrok provides and leave it running to prevent the URLs changing).

Run the command in the application's root directory:

 vonage apps:create "My Echo Server"  --voice_answer_url=https://example.com/webhooks/answer --voice_event_url=https://example.com/webhooks/event

Make a note of the Application ID returned by this command. It will also download your private.key file which the Voice API uses to authenticate your application.

Link the Voice API application to your Vonage number

Use the application ID to link your virtual number:

vonage apps:link --number=<NUMBER>

Install dependencies

In the application's parent directory and run npm install:

npm install

AWS setup

Complete the following steps:

  1. Create an AWS account and create an Adminstrator user. Make a note of your AWS key and secret, because you cannot retrieve the secret later on.
  2. Install and configure the AWS CLI
  3. Use the following AWS CLI command to create two new S3 buckets, one for the call audio and the other for the generated transcripts. These must be unique across S3. Ensure that the region supports the Amazon Transcribe API and Cloudwatch Events:
aws s3 mb s3://your-audio-bucket-name --region us-east-1
aws s3 mb s3://your-transcription-bucket-name --region us-east-1

Configure the environment

Copy example.env to .env and configure the following settings:

  • VONAGE_APPLICATION_ID: The Vonage Voice Application ID you created earlier
  • VONAGE_PRIVATE_KEY_FILE: The name of your private key file, which should be in the root of your application directory. E.g. private.key
  • OTHER_PHONE_NUMBER: Another phone number you can call to create a conversation
  • AWS_KEY: Your Amazon Web Services key
  • AWS_SECRET: Your Amazon Web Services secret
  • AWS_REGION: Your Amazon Web Services region, e.g. us-east-1
  • S3_PATH: Your path to S3 bucket storage, which should include the AWS_REGION, e.g. https://s3-us-east-1.amazonaws.com
  • S3_AUDIO_BUCKET_NAME: A uniquely-named S3 bucket which will contain call audio mp3 files
  • S3_TRANSCRIPTS_BUCKET_NAME=: A uniquely-named S3 bucket which will contain transcripts of the call audio

Deploy your serverless function

The serverless function is a Lambda that executes when a transcription job completes. It makes a POST request to your application's /webhooks/transcription endpoint.

Deploy this function:

cd transcribeReadyService
serverless deploy

Running the code

Local Install

  1. In the root directory of your application, execute:
node index.js

  1. Call your Vonage virtual number from one of your personal numbers. The other number you specified in the OTHER_PHONE_NUMBER setting should ring - answer it when it does.

  2. Speak a few words into each handset. When you're finished, disconnect both.

  3. Watch the console as the transcription job is being processed. If everything works properly, you should receive a notification that your job is complete and you should find the call audio file in your recordings directory and the corresponding transcript (in JSON format) in transcripts. Note how the transcript is split into channels, one for each device you used. The application parses the completed transcription and displays the result for each channel in the console.

Docker Compose

You can also run the code using Docker Compose using the following command.

docker-compose up

Adding more callers

If you have more than two numbers, you can add more callers to the conversation. Simply create a connect action for each in the /webhooks/answer NCCO and increase the number of channels in the record action accordingly.

Getting Help

We love to hear from you so if you have questions, comments or find a bug in the project, let us know! You can either:

Further Reading