start_speaker_enrollment_job

start_speaker_enrollment_job(**kwargs)

Starts a new batch speaker enrollment job using specified details.

See also: AWS API Documentation

Request Syntax

response = client.start_speaker_enrollment_job(
    ClientToken='string',
    DataAccessRoleArn='string',
    DomainId='string',
    EnrollmentConfig={
        'ExistingEnrollmentAction': 'SKIP'|'OVERWRITE',
        'FraudDetectionConfig': {
            'FraudDetectionAction': 'IGNORE'|'FAIL',
            'RiskThreshold': 123
        }
    },
    InputDataConfig={
        'S3Uri': 'string'
    },
    JobName='string',
    OutputDataConfig={
        'KmsKeyId': 'string',
        'S3Uri': 'string'
    }
)
Parameters
  • ClientToken (string) --

    The idempotency token for starting a new speaker enrollment Job. If not provided, Amazon Web Services SDK populates this field.

    This field is autopopulated if not provided.

  • DataAccessRoleArn (string) --

    [REQUIRED]

    The IAM role Amazon Resource Name (ARN) that grants Voice ID permissions to access customer's buckets to read the input manifest file and write the job output file. Refer to Batch enrollment using audio data from prior calls for the permissions needed in this role.

  • DomainId (string) --

    [REQUIRED]

    The identifier of the domain that contains the speaker enrollment job and in which the speakers are enrolled.

  • EnrollmentConfig (dict) --

    The enrollment config that contains details such as the action to take when a speaker is already enrolled in Voice ID or when a speaker is identified as a fraudster.

    • ExistingEnrollmentAction (string) --

      The action to take when the specified speaker is already enrolled in the specified domain. The default value is SKIP , which skips the enrollment for the existing speaker. Setting the value to OVERWRITE replaces the existing voice prints and enrollment audio stored for that speaker with new data generated from the latest audio.

    • FraudDetectionConfig (dict) --

      The fraud detection configuration to use for the speaker enrollment job.

      • FraudDetectionAction (string) --

        The action to take when the given speaker is flagged by the fraud detection system. The default value is FAIL , which fails the speaker enrollment. Changing this value to IGNORE results in the speaker being enrolled even if they are flagged by the fraud detection system.

      • RiskThreshold (integer) --

        Threshold value for determining whether the speaker is a high risk to be fraudulent. If the detected risk score calculated by Voice ID is greater than or equal to the threshold, the speaker is considered a fraudster.

  • InputDataConfig (dict) --

    [REQUIRED]

    The input data config containing the S3 location for the input manifest file that contains the list of speaker enrollment requests.

    • S3Uri (string) -- [REQUIRED]

      The S3 location for the input manifest file that contains the list of individual enrollment or registration job requests.

  • JobName (string) -- A name for your speaker enrollment job.
  • OutputDataConfig (dict) --

    [REQUIRED]

    The output data config containing the S3 location where Voice ID writes the job output file; you must also include a KMS key ID to encrypt the file.

    • KmsKeyId (string) --

      The identifier of the KMS key you want Voice ID to use to encrypt the output file of a speaker enrollment job/fraudster registration job.

    • S3Uri (string) -- [REQUIRED]

      The S3 path of the folder where Voice ID writes the job output file. It has a *.out extension. For example, if the input file name is input-file.json and the output folder path is s3://output-bucket/output-folder , the full output file path is s3://output-bucket/output-folder/job-Id/input-file.json.out .

Return type

dict

Returns

Response Syntax

{
    'Job': {
        'CreatedAt': datetime(2015, 1, 1),
        'DataAccessRoleArn': 'string',
        'DomainId': 'string',
        'EndedAt': datetime(2015, 1, 1),
        'EnrollmentConfig': {
            'ExistingEnrollmentAction': 'SKIP'|'OVERWRITE',
            'FraudDetectionConfig': {
                'FraudDetectionAction': 'IGNORE'|'FAIL',
                'RiskThreshold': 123
            }
        },
        'FailureDetails': {
            'Message': 'string',
            'StatusCode': 123
        },
        'InputDataConfig': {
            'S3Uri': 'string'
        },
        'JobId': 'string',
        'JobName': 'string',
        'JobProgress': {
            'PercentComplete': 123
        },
        'JobStatus': 'SUBMITTED'|'IN_PROGRESS'|'COMPLETED'|'COMPLETED_WITH_ERRORS'|'FAILED',
        'OutputDataConfig': {
            'KmsKeyId': 'string',
            'S3Uri': 'string'
        }
    }
}

Response Structure

  • (dict) --

    • Job (dict) --

      Details about the started speaker enrollment job.

      • CreatedAt (datetime) --

        A timestamp showing the creation of the speaker enrollment job.

      • DataAccessRoleArn (string) --

        The IAM role Amazon Resource Name (ARN) that grants Voice ID permissions to access customer's buckets to read the input manifest file and write the job output file.

      • DomainId (string) --

        The identifier of the domain that contains the speaker enrollment job.

      • EndedAt (datetime) --

        A timestamp showing when the speaker enrollment job ended.

      • EnrollmentConfig (dict) --

        The configuration that defines the action to take when the speaker is already enrolled in Voice ID, and the FraudDetectionConfig to use.

        • ExistingEnrollmentAction (string) --

          The action to take when the specified speaker is already enrolled in the specified domain. The default value is SKIP , which skips the enrollment for the existing speaker. Setting the value to OVERWRITE replaces the existing voice prints and enrollment audio stored for that speaker with new data generated from the latest audio.

        • FraudDetectionConfig (dict) --

          The fraud detection configuration to use for the speaker enrollment job.

          • FraudDetectionAction (string) --

            The action to take when the given speaker is flagged by the fraud detection system. The default value is FAIL , which fails the speaker enrollment. Changing this value to IGNORE results in the speaker being enrolled even if they are flagged by the fraud detection system.

          • RiskThreshold (integer) --

            Threshold value for determining whether the speaker is a high risk to be fraudulent. If the detected risk score calculated by Voice ID is greater than or equal to the threshold, the speaker is considered a fraudster.

      • FailureDetails (dict) --

        Contains details that are populated when an entire batch job fails. In cases of individual registration job failures, the batch job as a whole doesn't fail; it is completed with a JobStatus of COMPLETED_WITH_ERRORS . You can use the job output file to identify the individual registration requests that failed.

        • Message (string) --

          A description of the error that caused the batch job failure.

        • StatusCode (integer) --

          An HTTP status code representing the nature of the error.

      • InputDataConfig (dict) --

        The input data config containing an S3 URI for the input manifest file that contains the list of speaker enrollment job requests.

        • S3Uri (string) --

          The S3 location for the input manifest file that contains the list of individual enrollment or registration job requests.

      • JobId (string) --

        The service-generated identifier for the speaker enrollment job.

      • JobName (string) --

        The client-provided name for the speaker enrollment job.

      • JobProgress (dict) --

        Provides details on job progress. This field shows the completed percentage of registration requests listed in the input file.

        • PercentComplete (integer) --

          Shows the completed percentage of enrollment or registration requests listed in the input file.

      • JobStatus (string) --

        The current status of the speaker enrollment job.

      • OutputDataConfig (dict) --

        The output data config containing the S3 location where Voice ID writes the job output file; you must also include a KMS key ID to encrypt the file.

        • KmsKeyId (string) --

          The identifier of the KMS key you want Voice ID to use to encrypt the output file of a speaker enrollment job/fraudster registration job.

        • S3Uri (string) --

          The S3 path of the folder where Voice ID writes the job output file. It has a *.out extension. For example, if the input file name is input-file.json and the output folder path is s3://output-bucket/output-folder , the full output file path is s3://output-bucket/output-folder/job-Id/input-file.json.out .

Exceptions

  • VoiceID.Client.exceptions.ServiceQuotaExceededException
  • VoiceID.Client.exceptions.ResourceNotFoundException
  • VoiceID.Client.exceptions.ValidationException
  • VoiceID.Client.exceptions.ConflictException
  • VoiceID.Client.exceptions.InternalServerException
  • VoiceID.Client.exceptions.ThrottlingException
  • VoiceID.Client.exceptions.AccessDeniedException