안녕하세요 JeongU,
우리는 en-US-Davis:DragonHDLatestNeural 음성에 대한 귀하의 문제를 재현할 수 있으며, ssml 스크립트를 전달할 때 입력과 함께 추가적인 잡음이 발생하고 있습니다.
그래서 우리는 이것을 내부 제품 팀에 업데이트하고 세부 사항을 받는 대로 여러분께 업데이트하겠습니다.
하지만 SSML 스크립트 없이 일반 텍스트를 시도했을 때, 올바른 결과를 얻을 수 있었습니다.
그러므로 지금으로서는 아래 본문과 함께 일반 텍스트를 사용하는 우회 방법이 될 것입니다.
head = {
"Content-Type": "application/json",
"Ocp-Apim-Subscription-Key":"api-key"}
endpoint = "https://eastus.api.cognitive.microsoft.com/texttospeech/batchsyntheses/my-batch-synthesis-job-ssml-input-2?api-version=2024-04-01"
body = {
"inputKind": "PlainText",
"inputs": [
{
"content": 'HULA ZERO FOUR FIRE FOR EFFECT, and then.. QUEBEC SIERRA TWO FOUR FIFE, and then.. FIFE EIGHT ONE SEVEN SIX NINER SIX TWO K, and.. TWO TWO TUNGUSKA COMPANY HALTED IN OPEN'
}
],
"properties": {
"destinationContainerUrl": "destination_url",
"decompressOutputFiles": True
},
"synthesisConfig":{
"voice": "en-US-Davis:DragonHDLatestNeural",
"rate": "-3.51%",
"volume": "18.77%"
}
}
results = requests.put(endpoint,headers=head,data=json.dumps(body))
results.json()
아래에서 오디오 파일의 길이를 볼 수 있습니다.
문의 사항이 있으시면 댓글이나 개인 메시지로 알려주십시오.감사합니다