Guide-linked quiz · Hard Level
AI for Speech-Language Pathologists Quiz
Tests why AI word lists confuse spelling with phonemes, how ASR hides speech errors, and when automated articulation feedback can be trusted.
Question 1 of 8
Why might an AI list of '/k/ words' include 'knife' or 'city'?
Preview every question8 questions
Answers and explanations appear as you take the quiz.
- Why might an AI list of '/k/ words' include 'knife' or 'city'?
- The model was trained only on medical text
- Language models work with spelling rather than sounds
- Those words are common in children's books
- The model favors longer words
- A child says 'wabbit' and ASR transcribes 'rabbit.' What problem does this show?
- ASR cannot handle animal words
- The transcription is too slow
- The microphone was too far away
- ASR tends to turn errors into the intended word, which hides the error being measured
- According to the guide, what does Google's Project Relate aim to do?
- Generate minimal pair cards
- Score IEP goals
- Improve recognition of atypical speech by learning from the user's own recordings
- Replace language sample analysis
- Which prompt technique does the technical section recommend for more accurate word lists?
- Ask for an IPA transcription next to each word and specify position, syllable count and blends
- Ask for the words in alphabetical order
- Ask for twice as many words
- Use only one-syllable words
- Why should SLPs test an articulation app's automatic feedback against their own ears?
- A wrong 'correct' signal can reinforce errors
- Apps do not record audio
- ASHA bans automatic feedback
- The feedback only works in English
- For which speakers does the guide say ASR error rates rise sharply?
- Adult newscasters
- People reading from a script
- Professional voice actors
- Children and people with speech sound disorders, dysarthria or stuttering
- A child fronts /k/ to /t/. Which minimal pair from the guide fits?
- 'cat' and 'bat'
- 'key' and 'tea'
- 'sun' and 'fun'
- 'ship' and 'sip'
- What recording advice does the technical section give for speech analysis?
- Record in a busy room to reflect real life
- Delete the audio after transcribing, for privacy
- Record in a quiet room with the microphone at a consistent distance, and keep the original audio
- Always record through a phone speaker
Keep testing yourself
More quizzes picked for your level.