User Dictionary Registration

This API is used for importing and exporting registered words in the user dictionary.

Retrieving Registered Words from User Dictionary

You can retrieve a list of registered words in JSON format.

Endpoint

GET https://acp-api.amivoice.com/profilewords/<dgn>/<pid>?adf=<adf>

Authentication

Please specify the APPKEY in the Authorization header.

Authorization: Bearer {APPKEY}

Request Parameters

Parameter Name	Parameter Name	Notes
dgn	Connection Engine Name	Specify a connection engine name such as `-a-general`. For available values, please see List of Speech Recognition Engines. Set it as a path parameter.
pid	Profile ID	Any string. Set it as a path parameter.
adf	Audio Data Format	Specify either `16k` or `8k`. If not specified, it defaults to `16k`. Although some speech recognition engines support multiple sampling rates, user dictionary treats `8k` and `16k` as different engines, so you need to specify one. Set it as a query parameter. However, for End to End engines, `16k` and `8k` are treated as the same engine, so specification is not necessary.

Retrieving Profile List

You can retrieve a list of profiles in JSON format.

Endpoint

GET https://acp-api.amivoice.com/profilewords/<dgn>/?adf=<adf>

Authentication

Please specify the APPKEY in the Authorization header.

Authorization: Bearer {APPKEY}

Request Parameters

Parameter Name	Parameter Name	Note
dgn	Connection engine name	Specify the connection engine name such as `-a-general`. For available values, please see List of Speech Recognition Engines. Set it as a path parameter.
adf	Audio data format	Specify either `16k` or `8k`. If not specified, it defaults to `16k`. Although some speech recognition engines support multiple sampling rates, user dictionary is treated as different engines for `8k` and `16k`, so you need to specify one or the other. Set it as a query parameter. However, for End to End engines, `16k` and `8k` are treated as the same engine, so specification is not necessary.

Registering Words in User Dictionary

You can replace words in the user dictionary for the specified speech recognition engine and profile ID.

Endpoint

POST https://acp-api.amivoice.com/profilewords/<dgn>/<pid>?adf=<adf>

Authentication

Please specify the APPKEY in the Authorization header.

Authorization: Bearer {APPKEY}

Request Parameters

Parameter Name	Parameter Name	Notes
dgn	Connection Engine Name	Specify a connection engine name such as `-a-general`. For available values, please see List of Speech Recognition Engines. Set it as a path parameter.
pid	Profile ID	Any string. Set it as a path parameter.
adf	Audio Data Format	Specify either `16k` or `8k`. If not specified, it defaults to `16k`. Although some speech recognition engines support multiple sampling rates, user dictionary treats `8k` and `16k` as different engines, so you need to specify one. Set it as a query parameter. However, for End to End engines, `16k` and `8k` are treated as the same engine, so specification is not necessary.

Content Type

Content-Type: application/json

Request Body

For Hybrid Engines

Field Name	Description	Notes
`profilewords`	Word Registration Information	Specify the words you want to register in array format
`profilewords[].written`	Written	Specify the written form of the word
`profilewords[].spoken`	Pronunciation	Specify the pronunciation of the word
`profilewords[].classname`	Class Name	Specify the class name. This can be omitted if not set.

Example

{
    "profilewords": [
        {
            "written": "AMI",
            "spoken": "あみ",
        },
        {
            "written": "AOI",
            "spoken": "あおい",
        },
        {
            "written": "躑躅森",
            "spoken": "つつじもり",
            "classname": "名前"
        }
    ]
}

For End to End Engines

Field Name	Description	Notes
`profilewords`	Keyword Biasing Information	Specify the words you want to boost in array format
`profilewords[].written`	Written	Specify the written form of the word
`profilewords[].alternativewritten`	Alternative Written	Specify the written form of the word to be replaced by the "Written"
`profilewords[].biasing`	Biasing Level	Specify the strength of keyword biasing. Values from 0 to 1 can be specified, where 0 means no boost and 1 means strong boost at the cost of overall recognition accuracy. If not specified, it defaults to 0.5

It's possible to specify multiple alternative written for the same written. In this case, the same biasing level is applied to words with the same written. If different biasing levels are specified for the same written in the request, the value of the first occurrence in the request is adopted. In the following example, the biasing level for the same written "AMI" is uniformly applied with the value "0.8" which appears first. The biasing level for the third written "AOI" is applied with the default "0.5" as no specification is given.

Example

{
    "profilewords": [
        {
            "written": "AMI",
            "alternativewritten": "アドバンストメディア",
            "biasing": 0.8
        },
       {
            "written": "AMI",
            "alternativewritten": "あみ"
        },
        {
            "written": "AOI",
            "alternativewritten": "あおい"
        }
    ]
}

Response

On Success

HTTP Status Code: 200 OK

Response Body

For Hybrid Engines

Field Name	Description	Notes
`profilewords`	Word Registration Information	Registered words are obtained in array format
`profilewords[].written`	Written	The written form of the word is obtained
`profilewords[].spoken`	Pronunciation	The pronunciation of the word is obtained
`profilewords[].classname`	Class Name	The class name is returned. It is omitted if not set

Example

{
    "profilewords": [
        {
            "written": "AMI",
            "spoken": "あみ",
        },
        {
            "written": "AOI",
            "spoken": "あおい",
        },
        {
            "written": "躑躅森",
            "spoken": "つつじもり",	
            "classname": "名前"			
        }
    ]
}

For End to End Engines

Field Name	Description	Notes
`profilewords`	Keyword Biasing Information	The words to be boosted are returned in array format
`profilewords[].written`	Written	The written form of the word is returned
`profilewords[].alternativewritten`	Alternative Written	The written form of the word to be replaced by the "Written" is returned
`profilewords[].biasing`	Biasing Level	The strength of keyword biasing is returned. It is omitted when the value is the default 0.5

Example

{
    "profilewords": [
        {
            "written": "AMI",
            "alternativewritten": "あみ",
            "biasing": 0.8
        },
        {
            "written": "AMI",
            "alternativewritten": "アドバンストメディア",
            "biasing": 0.8
        },
        {
            "written": "AOI",
            "alternativewritten": "あおい"
        }
    ]
}

On Failure

HTTP Status Code: Other than 200

Response Body

Field Name	Description	Notes
code	Error Code	Code indicating the type of error
message	Error Message	Text explaining the details of the error content

Error Codes and Error Messages

HTTP Status Code	Response Body	Error Situation
404 Not Found	`{"code":"?","message":"missing dictation grammar file name"}`	When the dictation grammar file name `<dgn>` is not specified
404 Not Found	`{"code":"?","message":"missing profile id"}`	When the profile ID `<pid>` is not specified
405 Method Not Allowed	`{"code":"?","message":"POST method is not allowed"}`	When POST is specified
400 Bad Request	`{"code":"-","message":"missing request body"}`	When a POST method is requested but no request body is specified
400 Bad Request	`{"code":"-","message":"invalid request body (<optionalMessage>)"}`	When the request body is not valid JSON
422 Unprocessable Entity	`{"code":"?","message":"missing profile words"}`	When the JSON specified in the request body is not in the expected format
422 Unprocessable Entity	`{"code":"?","message":"invalid written: <written>"}`	When the written in the JSON specified in the request body is incorrect
422 Unprocessable Entity	`{"code":"?","message":"invalid spoken: <spoken>"}`	When the pronunciation in the JSON specified in the request body is incorrect
422 Unprocessable Entity	`{"code":"?","message":"invalid className: <className>"}`	When the word class in the JSON specified in the request body is incorrect
422 Unprocessable Entity	`{"code":"?","message":"invalid alternative written: <alternativeWritten>"}`	When the alternative written in the JSON specified in the request body is incorrect
401 Unauthorized	`{"code":"-","message":"can't verify service authorization"}`	When authentication of the APPKEY (service authentication key string) fails
401 Unauthorized	`{"code":"-","message":"service authorization has expired: <expirationTime> (<expiresIn>)"}`	When authentication of the APPKEY (service authentication key string) fails
500 Internal Server Error	`{"code":"?","message":"unexpected exception occurred (<optionalMessage>)"}`	When an unexpected exception occurs

Example

{
    "code":"?",
    "message":"missing dictation grammar file name"
}

Retrieving Registered Words from User Dictionary​

Endpoint​

Authentication​

Request Parameters​

Retrieving Profile List​

Endpoint​

Authentication​

Request Parameters​

Registering Words in User Dictionary​

Endpoint​

Authentication​

Request Parameters​

For Hybrid Engines​

For End to End Engines​

Response​

On Success​

For Hybrid Engines​

For End to End Engines​

On Failure​

Retrieving Registered Words from User Dictionary

Endpoint

Authentication

Request Parameters

Retrieving Profile List

Endpoint

Authentication

Request Parameters

Registering Words in User Dictionary

Endpoint

Authentication

Request Parameters

For Hybrid Engines

For End to End Engines

Response

On Success

For Hybrid Engines

For End to End Engines

On Failure