Audio Available

Do we have an audio file for an article?

Currently hosted on Heroku.

What does it do?

audio-available is the place to find out whether or not there is an audio version of an article on FT.com.

Why does it exist?

At the moment, audio content doesn't have a proper, permanent place to live at the FT. Over the last little while, FT Labs has been exploring different applications of audio across the FT's content. Having a simple service that can tell us whether or not a given article has an audio file, and where we can find it, seemed like a pretty useful thing to have.

Usage

Checking for an audio article

To check for the existence of an audio file for an article, you can hit the /check/:UUID.

You can pass any FT content UUID to check the availability, location and basic information of an audio file.

The UUID must be a valid v4 UUID.

Example

https://audio-available.ft.com/check/0e69c5a4-a4cd-11e6-8898-79a99e2a4de6

Response

The HTTP status code for any lookup will be 200, even if a file does not exist. A status code of 420 will be returned if the request is improper (invalid UUID, for example).

// If file exists...

{
	"haveFile": true,
	"url": "[ABSOLUTE PATH TO AUDIO FILE]",
	provider: "ftlabs-tts",
	provider_name: "FT Labs text-to-speech service",
	ishuman : "true"
	"size": "2744769",
	"duration": {
		"milliseconds": 343040,
		"seconds": "343.04",
		"humantime": "05:43"
	}
}

// If file does not exist...

{
	"haveFile": false
}

Triggering a purge of the audio-available cache

Caching has been moved from the app to Fastly. Every audio service (absorber/management et al) that can manipulate the S3 audio buckets or DynamoDB tables should trigger a cache purge to the audio available service. This can be done by making a PURGE request to the URL that is to be purged.

curl -X PURGE https://audio-available.ft.com/check/2cc51074-2c10-11e7-9ec8-168383da43b7

Response

{
  "status":"ok",
  "id":"108-1391560174-974124"
}

/check response properties

haveFile: [BOOLEAN] true or false

The following properties will only appear if there is a valid audio file

url: [STRING]

size: [NUMBER] size of available audio file in bytes

provider [STRING] The short name of any audio provider as determined by the config in the Absorber service

provider_name [STRING] The long version of the provider short name. This value is intended to be displayed to a user as audio is being played.

ishuman [BOOLEAN] Some audio files will be of generated voices. This property indicates whether or not the file has been recorded by a person or not

duration: [OBJECT]

• milliseconds: [NUMBER] length of audio in milliseconds
• seconds: [STRING] length of audio in seconds
• humantime: [STRING] human readable length of audio. Will be hh:mm:ss if longer than 1 hour, otherwise mm:ss.

Running

To build and run this service for yourself, a number of environment variables need to be set. These can be set by including a .env file in the root of the project directory.

Required

AWS_ACCESS_KEY_ID

• The access key ID for the user with Amazon DynamoDB/S3 read/write permissions

AWS_SECRET_ACCESS_KEY

• The corresponding secret access key for the AWS_ACCESS_KEY_ID

AWS_AUDIO_BUCKET

• The name of the S3 bucket that the audio versions of FT articles can be found in.

AWS_AUDIO_METADATA_TABLE

• The name of the DynamoDB table that contains the metadata for the audio versions of FT articles.

CACHE_PURGE_KEY

• The value that the purgeToken needs to be to trigger a cache purge at the /purge endpoint.

Optional

SL_MEDIA_FORMAT

• The expected media format for the audio files. Defaults to .mp3 if not included

AWS_REGION

• The region code for the AWS services being utilised. Defaults to us-west-2 if not included.

HOLDING_TIME

• How long a request for an audio file lookup should be held before returning an error/result. A request will only be held if the first check for an audio file is underway.

Additional info

This service has /__gtg and /__health endpoints.