Getting an access token

The token is issued for:

• recognition of audio files;
• recognition of streaming audio;
• speech synthesis.

There are two ways to get a token:

1. OAuth is the preferred scenario as it implements a refresh token mechanism.

An example of a request to generate a token:

Copy
curl -X POST --location 'https://mcs.mail.ru/auth/oauth/v1/token' \--header 'Content-Type: application/json' \--data '{"client_id":"mcs5719255887.voice.6Wg9D18XPZN28DWy4wEba","client_secret": "1234567890ABCDEFGHIJKLMNOPQRSTUVWXYZ","grant_type":"client_credentials"}'

Answer example:

Copy
{  "refresh_token": "ABCDEFGHIJKLMNOPQRSTUVWXYZ1234567890",  "access_token": "0987654321ABCDEFGHIJKLMNOPQRSTUVWXYZ",  "expired_in": "3600",  "scope": {    "tts": 1,    "asr_short": 1,    "asr_stream": 1  }}

In response from the server, you will receive 2 tokens:

• access_token — access token (becomes invalid in an hour);
• refresh_token — token for refreshing the access token.

Example of a token refresh request:

Copy
curl -X POST --location 'https://mcs.mail.ru/auth/oauth/v1/token' \--header 'Content-Type: application/json' \--data '{"client_id":"mcs5719255887.voice.6Wg9D18XPZN28DWy4wEba","refresh_token":"ABCDEFGHIJKLMNOPQRSTUVWXYZ1234567890","grant_type":"refresh_token"}'

Answer example:

Copy
{  "refresh_token": "ABCDEFGHIJKLMNOPQRSTUVWXYZ1234567890",  "access_token": "23wcTiUeaa334Uv6TH3StRGiZBMCyyo1Hzgqqa3n",  "expired_in": "3600",  "scope": {    "tts": 1,    "asr_short": 1,    "asr_stream": 1  }}

A project can have up to 25 active refresh_token at a time, each of them can have 25 access_token associated at any given time.

2. Service Token is a less preferred scenario. The token is issued once and can work without time limit. The only risk of using such a token is that when the token is selected or compromised, an attacker may be able to use the service in very large volumes, which will entail unexpected expenses. When the funds on the personal account are exhausted, the service will be suspended.

To get a service token, click Machine Learning → Voice API → Add Service Token.

Select the required checkboxes and click "Create".

The received access token must be sent in all HTTP requests in the Bearer Token format.

Speech recognition

Speech synthesis