Bitzu

AI-901 ยท Common AI workloads

What can Azure Speech do, and which mode fits which job?

Speech-to-text vs text-to-speech, real-time vs batch transcription, voices and neural speech, speech translation, and typical scenarios such as captions and voice assistants.

Key points

  • Speech to text turns audio into text:
  • Real-time transcription: streaming audio (live captions, voice commands).
  • Fast transcription: pre-recorded files.
  • Batch transcription: large volumes, asynchronous.
  • Text to speech: text into natural neural voices; SSML fine-tunes pitch, rate and pronunciation.
  • Speech translation: spoken language into another language as text or speech.
  • Custom speech: helps with noise and jargon.

Source: https://learn.microsoft.com/en-us/azure/ai-services/speech-service/overview

Practice questions

Single choice

Which capability converts written text into spoken audio?

  1. Speech to text
  2. Text to speech
  3. Optical character recognition
  4. Key phrase extraction
Show answer

Answer: B. Text to speech

Text to speech (speech synthesis) produces audio from text. Speech to text does the reverse, and OCR reads text from images.

Select all that apply

Which statements about Azure Speech are correct?

  1. Neural voices are powered by deep neural networks
  2. SSML can adjust pitch, speaking rate and pronunciation
  3. Custom speech models can help with noisy audio and domain jargon
  4. Batch transcription is designed for live streaming audio
  5. Speech to text outputs spoken audio
Show answer

Answer: A. Neural voices are powered by deep neural networks, B. SSML can adjust pitch, speaking rate and pronunciation, C. Custom speech models can help with noisy audio and domain jargon

Neural voices are powered by deep neural networks, SSML adjusts pitch, rate and pronunciation, and custom speech models help with noise and jargon. Batch transcription is for asynchronous processing, not live streams.

True or false

True or false: speech translation can only produce written text, never spoken output.

Show answer

Answer: False

False. Speech translation supports both speech-to-text and speech-to-speech translation.

Practice 8 questions on this topic in Bitzu

Short lessons, instant explanations and spaced review of what you get wrong. Free on iOS and Android.

Download on the App StoreGet it on Google Play

More AI-901 topics in this module