1回答
- 新しい順
- 投票が多い順
- コメントが多い順
0
Hi,
Amazon Transcribe can build custom language models using recorded voice data. It can also do speaker diarization to partition different speakers. Details of its usage can refer AWS official document
Hope it helps.
回答済み 1年前
関連するコンテンツ
- AWS公式更新しました 2年前
- AWS公式更新しました 2年前
Thanks for the answer but it doesn't help me much, I already use the "speaker diarization" functionality, and I get a transcript where each one of the speakers that appear in the audio is labeled, what I need now is to know if there is any way to instead of obtaining a label like this to identify the speaker "spk_0", if I can save the voice of that person so that the amazon transcription service recognizes it by its tone of voice and that when the transcript is generated, in Instead of having a generic label that says "spk_0", obtain a personalized label that indicates the name of the person to whom that tone of voice is associated, for example "Carlos" or "Pedro".
Thanks for your further explanation and I can have a better understanding of your question now.
This is a very practical scenario which is not current supported by Amazon Transcribe alone as a single feature. But you still have several options to do.
One direction is to have a two-step solution by using Amazon Transcribe as the first step. You may implement your personal voice authentication using SageMaker.
Another direction is that if you are facing a cloud contact center case or with similar using situations, Amazon Connect is another SaaS solution for you to consider. Amazon Connect Voice ID uses machine learning to verify the identity of genuine customers by analyzing a caller's unique voice characteristics. For more details, you may refer its official document.
Hope it helps.