AI-102 Question 244
Select 3You are designing a real-time speech-to-speech translation solution for an international conference using Azure AI Speech service. The solution must translate spoken English into French and play the translated audio back to the audience in near real-time. Which components must be included in your architecture to achieve this?
- A
Speech-to-Text service to convert spoken English into text
- B
Text Translation service to translate the English text into French
- C
Text-to-Speech service to synthesize French audio from translated text
- D
Speech Synthesis Markup Language (SSML) for customizing voice output
- E
Form Recognizer service to extract structured data from the speech input
Show answer and explanation
Correct answers: A, B, C
Explanation
Speech-to-speech translation involves three key stages: converting spoken input into text (Speech-to-Text), translating the text into the target language (Text Translation), and synthesizing audio output from the translated text (Text-to-Speech). While additional features like SSML can enhance the solution, they are not strictly required for the core functionality. Form Recognizer is unrelated to this use case.
- A. Correct.
Speech-to-Text is necessary to transcribe the spoken English into text, which is the first step in the speech-to-speech translation process.
- B. Correct.
Text Translation is needed to translate the transcribed English text into French, enabling the language conversion.
- C. Correct.
Text-to-Speech is required to synthesize the translated French text into audio that can be played back to the audience.
- D. Incorrect.
SSML is useful for customizing voice output, such as tone or pitch, but it is not a mandatory component for basic speech-to-speech translation.
- E. Incorrect.
Form Recognizer is used for extracting structured data from documents or forms and is not related to speech-to-speech translation.