AWS re:Postを使用することにより、以下に同意したことになります AWS re:Post 利用規約
AWS re:Postre:Post

Speaker partitioning overlap times

1

In the documentation for speaker partitioning, it mentions:

Utterances that overlap in the input audio don't overlap in the transcription output.

Does this mean that the transcribed text itself does not overlap? Or will the times never overlap?

For example, if Speaker A has times 1 -> 5 and Speaker B has times 3 -> 7, how will these times appear in the transcription output?

In tests, I have not been able to see these times represented accurately with overlap. Is it possible to change settings to allow for this?

トピック
機械学習と AI
タグ
Amazon Transcribe
言語
English
Austin
質問済み 10ヶ月前224ビュー
1回答
0

Hi,

The documentation says

If an utterance from one speaker overlaps with an utterance from another speaker, Amazon Transcribe 
Medical orders them in the transcription by their start times. Utterances that overlap in the input audio 
don't overlap in the transcription output.

So, the ones who starts speaking first by start time is transcribed first for his full utterance and the comes the ones who started second.

Best,

Didier

profile pictureAWS
エキスパート
Didier_Durand
回答済み 10ヶ月前
  • Austin
    10ヶ月前

    But in the example of "Speaker A has times 1 -> 5 and Speaker B has times 3 -> 7," would that not still match that description? The results we're seeing are instead Speaker A 1 -> 5 and Speaker B 5 -> 7 which is not accurate to the input audio. Just making sure we don't have an improper configuration.

ログインしていません。 ログイン 回答を投稿する。

優れた回答とは、質問に明確に答え、建設的なフィードバックを提供し、質問者の専門分野におけるスキルの向上を促すものです。

質問に答えるためのガイドライン

関連するコンテンツ