Use CreateVocabulary com um AWS SDK ou CLI

    /// <summary>
    /// Create a custom vocabulary using a list of phrases. Custom vocabularies
    /// improve transcription accuracy for one or more specific words.
    /// </summary>
    /// <param name="languageCode">The language code of the vocabulary.</param>
    /// <param name="phrases">Phrases to use in the vocabulary.</param>
    /// <param name="vocabularyName">Name for the vocabulary.</param>
    /// <returns>The state of the custom vocabulary.</returns>
    public async Task<VocabularyState> CreateCustomVocabulary(LanguageCode languageCode,
        List<string> phrases, string vocabularyName)
    {
        var response = await _amazonTranscribeService.CreateVocabularyAsync(
            new CreateVocabularyRequest
            {
                LanguageCode = languageCode,
                Phrases = phrases,
                VocabularyName = vocabularyName
            });
        return response.VocabularyState;
    }

Como criar um vocabulário personalizado

O exemplo de create-vocabulary a seguir cria um vocabulário personalizado. Para criar um vocabulário personalizado, você deve ter criado um arquivo de texto com todos os termos que deseja transcrever com mais precisão. Para vocabulary-file-uri, especifique o Amazon Simple Storage Service (Amazon URI S3) desse arquivo de texto. Em language-code, especifique um código de idioma correspondente ao idioma do vocabulário personalizado. Em vocabulary-name, especifique como você deseja chamar seu vocabulário personalizado.

aws transcribe create-vocabulary \
    --language-code language-code \
    --vocabulary-name cli-vocab-example \
    --vocabulary-file-uri s3://DOC-EXAMPLE-BUCKET/Amazon-S3-prefix/the-text-file-for-the-custom-vocabulary.txt

Saída:

{
    "VocabularyName": "cli-vocab-example",
    "LanguageCode": "language-code",
    "VocabularyState": "PENDING"
}

Para obter mais informações, consulte Custom Vocabularies no Guia do desenvolvedor do Amazon Transcribe.

def create_vocabulary(
    vocabulary_name, language_code, transcribe_client, phrases=None, table_uri=None
):
    """
    Creates a custom vocabulary that can be used to improve the accuracy of
    transcription jobs. This function returns as soon as the vocabulary processing
    is started. Call get_vocabulary to get the current status of the vocabulary.
    The vocabulary is ready to use when its status is 'READY'.

    :param vocabulary_name: The name of the custom vocabulary.
    :param language_code: The language code of the vocabulary.
                          For example, en-US or nl-NL.
    :param transcribe_client: The Boto3 Transcribe client.
    :param phrases: A list of comma-separated phrases to include in the vocabulary.
    :param table_uri: A table of phrases and pronunciation hints to include in the
                      vocabulary.
    :return: Information about the newly created vocabulary.
    """
    try:
        vocab_args = {"VocabularyName": vocabulary_name, "LanguageCode": language_code}
        if phrases is not None:
            vocab_args["Phrases"] = phrases
        elif table_uri is not None:
            vocab_args["VocabularyFileUri"] = table_uri
        response = transcribe_client.create_vocabulary(**vocab_args)
        logger.info("Created custom vocabulary %s.", response["VocabularyName"])
    except ClientError:
        logger.exception("Couldn't create custom vocabulary %s.", vocabulary_name)
        raise
    else:
        return response