다음을 통해 공유


Models - Create

새 음성 모델을 만듭니다.

PUT {endpoint}/customvoice/models/{id}?api-version=2024-02-01-preview

URI 매개 변수

Name In(다음 안에) 필수 형식 Description
endpoint
path True

string

지원되는 Cognitive Services 엔드포인트(프로토콜 및 호스트 이름( 예: https://eastus.api.cognitive.microsoft.com).

id
path True

string

리소스의 ID입니다.

regex 패턴: ^[a-zA-Z0-9][a-zA-Z0-9._-]{1,62}[a-zA-Z0-9]$

api-version
query True

string

이 작업에 사용할 API 버전입니다.

요청 헤더

Name 필수 형식 Description
Ocp-Apim-Subscription-Key True

string

여기에서 Speech 리소스 키를 제공합니다.

Operation-Id

string

작업에 대한 상태 모니터의 ID입니다. Operation-Id 헤더가 기존 작업과 일치하고 요청이 이전 요청과 동일하지 않으면 400 잘못된 요청으로 실패합니다.

regex 패턴: ^[a-zA-Z0-9][a-zA-Z0-9._-]{1,62}[a-zA-Z0-9]$

요청 본문

Name 필수 형식 Description
consentId True

string

리소스 ID

projectId True

string

리소스 ID

recipe True

Recipe

모델 빌드를 위한 레시피입니다. 레시피가 다르면 기능이 다릅니다.

trainingSetId True

string

리소스 ID

description

string

모델 설명

id

string

리소스 ID

locale

string

이 모델의 로캘입니다. 로캘 코드는 BCP-47을 따릅니다. 텍스트 음성 변환 로캘 목록은 여기에서 찾을 수 있습니다 https://learn.microsoft.com/azure/ai-services/speech-service/language-support?tabs=tts.

properties

ModelProperties

모델 속성

status

Status

리소스의 상태입니다.

voiceName

string

음성 이름

응답

Name 형식 Description
201 Created

Model

생성일

헤더

  • Operation-Id: string
  • Operation-Location: string
Other Status Codes

ErrorResponse

오류가 발생했습니다.

헤더

x-ms-error-code: string

보안

Ocp-Apim-Subscription-Key

여기에서 Speech 리소스 키를 제공합니다.

형식: apiKey
In(다음 안에): header

예제

Create a model
Create a multi style model

Create a model

샘플 요청

PUT {endpoint}/customvoice/models/Jessica?api-version=2024-02-01-preview



{
  "voiceName": "JessicaNeural",
  "description": "Jessica voice",
  "recipe": {
    "kind": "Default"
  },
  "projectId": "Jessica",
  "consentId": "Jessica",
  "trainingSetId": "Jessica-300"
}

샘플 응답

Operation-Location: https://eastus.api.cognitive.microsoft.com/customvoice/operations/1f4352df-f247-40c0-a7b1-a54d017933e1?api-version=2024-02-01-preview
Operation-Id: 1f4352df-f247-40c0-a7b1-a54d017933e1
{
  "id": "Jessica",
  "voiceName": "JessicaNeural",
  "description": "Jessica voice",
  "recipe": {
    "kind": "Default",
    "version": "V7.2023.03"
  },
  "projectId": "Jessica",
  "consentId": "Jessica",
  "trainingSetId": "Jessica-300",
  "locale": "en-US",
  "engineVersion": "2023.07.04.0",
  "status": "NotStarted",
  "createdDateTime": "2023-04-01T05:30:00.000Z",
  "lastActionDateTime": "2023-04-02T10:15:30.000Z"
}

Create a multi style model

샘플 요청

PUT {endpoint}/customvoice/models/JessicaMultiStyle?api-version=2024-02-01-preview



{
  "voiceName": "JessicaMultiStyleNeural",
  "description": "Jessica multi style voice",
  "recipe": {
    "kind": "MultiStyle"
  },
  "projectId": "Jessica",
  "consentId": "Jessica",
  "trainingSetId": "Jessica-300",
  "locale": "en-US",
  "properties": {
    "presetStyles": [
      "cheerful",
      "sad"
    ],
    "styleTrainingSetIds": {
      "happy": "JessicaHappy-300",
      "myStyle2": "JessicaStyle2"
    }
  }
}

샘플 응답

Operation-Location: https://eastus.api.cognitive.microsoft.com/customvoice/operations/a01a127a-c204-4e46-a8c1-fab01559b05b?api-version=2024-02-01-preview
Operation-Id: a01a127a-c204-4e46-a8c1-fab01559b05b
{
  "id": "JessicaMultiStyle",
  "voiceName": "JessicaMultiStyleNeural",
  "description": "Jessica multi style voice",
  "recipe": {
    "kind": "MultiStyle",
    "version": "V3.2023.06"
  },
  "projectId": "Jessica",
  "consentId": "Jessica",
  "trainingSetId": "Jessica-300",
  "locale": "en-US",
  "engineVersion": "2023.07.04.0",
  "properties": {
    "presetStyles": [
      "cheerful",
      "sad"
    ],
    "styleTrainingSetIds": {
      "happy": "JessicaHappy-300",
      "myStyle2": "JessicaStyle2"
    },
    "voiceStyles": [
      "cheerful",
      "sad",
      "happy",
      "myStyle2"
    ]
  },
  "status": "NotStarted",
  "createdDateTime": "2023-04-01T05:30:00.000Z",
  "lastActionDateTime": "2023-04-02T10:15:30.000Z"
}

정의

Name Description
Error

최상위 오류는 에서 https://github.com/microsoft/api-guidelines/blob/vNext/azure/Guidelines.md#handling-errors사용할 수 있는 Microsoft Azure REST API 지침을 따릅니다. 여기에는 오류 코드, 메시지, 세부 정보, 대상 및 보다 설명적인 세부 정보가 포함된 내부 오류가 포함된 최상위 오류가 포함됩니다.

ErrorCode

최상위 오류 코드

ErrorResponse

오류 응답은 에서 https://github.com/microsoft/api-guidelines/blob/vNext/azure/Guidelines.md#handling-errors사용할 수 있는 Microsoft Azure REST API 지침을 따릅니다.

InnerError

내부 오류는 에서 https://github.com/microsoft/api-guidelines/blob/vNext/azure/Guidelines.md#handling-errors사용할 수 있는 Microsoft Azure REST API 지침을 따릅니다. 여기에는 필수 속성 오류 코드, 메시지 및 선택적 속성 대상, 내부 오류(중첩될 수 있음)가 포함됩니다.

Model

모델 개체

ModelFailureReason

모델 학습 실패 이유

ModelProperties

모델 속성

PresetStyleItem

레시피에서 지원하는 사전 설정 스타일입니다. 음성 모델은 스타일 학습 집합 없이 이러한 스타일을 지원할 수 있습니다.

Recipe

모델 빌드를 위한 레시피입니다. 레시피가 다르면 기능이 다릅니다.

Status

리소스의 상태입니다.

Error

최상위 오류는 에서 https://github.com/microsoft/api-guidelines/blob/vNext/azure/Guidelines.md#handling-errors사용할 수 있는 Microsoft Azure REST API 지침을 따릅니다. 여기에는 오류 코드, 메시지, 세부 정보, 대상 및 보다 설명적인 세부 정보가 포함된 내부 오류가 포함된 최상위 오류가 포함됩니다.

Name 형식 Description
code

ErrorCode

최상위 오류 코드

details

Error[]

오류 및/또는 예상 정책에 대한 추가 지원 세부 정보입니다.

innererror

InnerError

내부 오류는 에서 https://github.com/microsoft/api-guidelines/blob/vNext/azure/Guidelines.md#handling-errors사용할 수 있는 Microsoft Azure REST API 지침을 따릅니다. 여기에는 필수 속성 오류 코드, 메시지 및 선택적 속성 대상, 내부 오류(중첩될 수 있음)가 포함됩니다.

message

string

최상위 오류 메시지입니다.

target

string

오류의 출처입니다. 예를 들어 잘못된 모델의 경우 "model" 또는 "model id"가 됩니다.

ErrorCode

최상위 오류 코드

Name 형식 Description
BadArgument

string

BadRequest

string

Forbidden

string

InternalServerError

string

NotFound

string

ServiceUnavailable

string

TooManyRequests

string

Unauthorized

string

UnsupportedMediaType

string

ErrorResponse

오류 응답은 에서 https://github.com/microsoft/api-guidelines/blob/vNext/azure/Guidelines.md#handling-errors사용할 수 있는 Microsoft Azure REST API 지침을 따릅니다.

Name 형식 Description
error

Error

최상위 오류는 에서 https://github.com/microsoft/api-guidelines/blob/vNext/azure/Guidelines.md#handling-errors사용할 수 있는 Microsoft Azure REST API 지침을 따릅니다. 여기에는 오류 코드, 메시지, 세부 정보, 대상 및 보다 설명적인 세부 정보가 포함된 내부 오류가 포함된 최상위 오류가 포함됩니다.

InnerError

내부 오류는 에서 https://github.com/microsoft/api-guidelines/blob/vNext/azure/Guidelines.md#handling-errors사용할 수 있는 Microsoft Azure REST API 지침을 따릅니다. 여기에는 필수 속성 오류 코드, 메시지 및 선택적 속성 대상, 내부 오류(중첩될 수 있음)가 포함됩니다.

Name 형식 Description
code

string

진단에 도움이 되는 자세한 오류 코드입니다.

innererror

InnerError

내부 오류는 에서 https://github.com/microsoft/api-guidelines/blob/vNext/azure/Guidelines.md#handling-errors사용할 수 있는 Microsoft Azure REST API 지침을 따릅니다. 여기에는 필수 속성 오류 코드, 메시지 및 선택적 속성 대상, 내부 오류(중첩될 수 있음)가 포함됩니다.

message

string

자세한 오류 메시지입니다.

target

string

오류의 출처입니다. 예를 들어 잘못된 모델의 경우 "model" 또는 "model id"가 됩니다.

Model

모델 개체

Name 형식 Description
consentId

string

리소스 ID

createdDateTime

string

개체를 만들 때의 타임스탬프입니다. 타임스탬프는 ISO 8601 날짜 및 시간 형식으로 인코딩됩니다("YYYY-MM-DDThh:mm:ssZ", 참조 https://en.wikipedia.org/wiki/ISO_8601#Combined_date_and_time_representations).

description

string

모델 설명

engineVersion

string

엔진 버전입니다. 이 버전을 업데이트하면 최신 발음 버그 수정을 가져올 수 있습니다.

id

string

리소스 ID

lastActionDateTime

string

현재 상태가 입력된 타임스탬프입니다. 타임스탬프는 ISO 8601 날짜 및 시간 형식으로 인코딩됩니다("YYYY-MM-DDThh:mm:ssZ", 참조 https://en.wikipedia.org/wiki/ISO_8601#Combined_date_and_time_representations).

locale

string

이 모델의 로캘입니다. 로캘 코드는 BCP-47을 따릅니다. 텍스트 음성 변환 로캘 목록은 여기에서 찾을 수 있습니다 https://learn.microsoft.com/azure/ai-services/speech-service/language-support?tabs=tts.

projectId

string

리소스 ID

properties

ModelProperties

모델 속성

recipe

Recipe

모델 빌드를 위한 레시피입니다. 레시피가 다르면 기능이 다릅니다.

status

Status

리소스의 상태입니다.

trainingSetId

string

리소스 ID

voiceName

string

음성 이름

ModelFailureReason

모델 학습 실패 이유

Name 형식 Description
InaccessibleCustomerStorage

string

고객은 음성 계정에서 Bring Your Own Storage를 사용합니다. 하지만 지금은 스토리지에 액세스할 수 없습니다. 문서를 확인하세요.

Internal

string

사용자 지정 음성 서비스 오류입니다.

SpeakerVerificationFailed

string

동의 및 학습 오디오는 동일한 스피커에서 제공되지 않습니다.

TerminateByUser

string

고객이 모델 학습을 취소했습니다.

ModelProperties

모델 속성

Name 형식 Description
failureReason

ModelFailureReason

모델 학습 실패 이유

presetStyles

string[]

이 모델의 사전 설정 스타일입니다.

styleTrainingSetIds

object

사용자 지정된 스타일 및 관련 학습 집합.

voiceStyles

string[]

이 모델에서 지원하는 모든 스타일입니다.

PresetStyleItem

레시피에서 지원하는 사전 설정 스타일입니다. 음성 모델은 스타일 학습 집합 없이 이러한 스타일을 지원할 수 있습니다.

Name 형식 Description
female

string[]

여성 음성 모델에서 지원되는 사전 설정 스타일입니다.

male

string[]

남성 음성 모델에서 지원되는 사전 설정 스타일입니다.

Recipe

모델 빌드를 위한 레시피입니다. 레시피가 다르면 기능이 다릅니다.

Name 형식 Description
datasetLocales

string[]

학습 데이터 세트의 로캘입니다. 로캘 코드는 BCP-47을 따릅니다. 텍스트 음성 변환 로캘 목록은 여기에서 찾을 수 있습니다 https://learn.microsoft.com/azure/ai-services/speech-service/language-support?tabs=tts.

description

string

레시피 설명

kind

string

레시피 종류

maxCustomStyleNum

integer

한 음성 모델에서 지원되는 최대 사용자 지정 스타일 번호입니다.

minStyleUtteranceCount

integer

사용자 지정된 각 스타일을 학습하는 데 필요한 최소 발화 수입니다.

minUtteranceCount

integer

이 레시피를 사용하여 음성 모델을 학습시키는 데 필요한 최소 발화 수입니다.

modelLocales

string[]

음성 모델이 이 레시피로 말할 수 있는 로캘입니다. 로캘 코드는 BCP-47을 따릅니다. 텍스트 음성 변환 로캘 목록은 여기에서 찾을 수 있습니다 https://learn.microsoft.com/azure/ai-services/speech-service/language-support?tabs=tts.

presetStyles

<string,  PresetStyleItem>

로캘당 이 레시피에서 지원하는 사전 설정 스타일입니다. 스타일 학습 집합 없이 이러한 스타일을 가져올 수 있습니다.

version

string

레시피 버전

Status

리소스의 상태입니다.

Name 형식 Description
Disabled

string

Disabling

string

Failed

string

NotStarted

string

Running

string

Succeeded

string