Kedis ku xidhan hagaha · Dhexdhexaad Heerka

Kediska Aqoonsiga Dareenka Hadalka

Waxay hubisaa dhawaaqyada dhawaaqa, xaddidaadaha xogta, guud ahaan isu-tagga jirka, calaamadaynta madmadowga, iyo goobta shaqada ee EU-da iyo mamnuucida waxbarashada.

Wadooyinka haga ee la xidhiidhaAqoonsiga Dareenka Hadalka
Su'aal 1 ee 8

Muxuu ugu horrayn u falanqeeyaa Aqoonsiga Dareenka Hadalka?

Horudhac su'aal kasta8 su'aalo

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

  1. Muxuu ugu horrayn u falanqeeyaa Aqoonsiga Dareenka Hadalka?
    • Astaamaha acoustic ee codka sida garka iyo tamarta
    • Erayada dhabta ah ee qoraalka
    • Naxwaha weedha
    • Hadalka wejigiisa
  2. Which acoustic properties can speech-emotion models use as cues?
    • Only the spelling of transcribed words
    • Pitch, energy, speaking rate, and voice-quality features
    • A person’s job title
    • The number of speakers in a company
  3. Waa maxay sababta calaamadaynta xogta shucuurta ay gaar ahaan u adag tahay?
    • Galalka maqalku aad bay u weyn yihiin in la kaydiyo
    • Makarafoonadu ma duubi karaan shucuurta
    • Shucuurtu waa mid shakhsi ahaaneed, tafsiirayaasha aadamuhuna inta badan way khilaafaan
    • Qiirodu weligood kuma muuqdaan hadalka
  4. Midkee ayaa ah xog-ururinta hadalka shucuurta ee caanka ah?
    • ImageNet
    • COCO
    • SQAD
    • IEMOCAP
  5. Sidee hababka casriga ah ee SER inta badan uga faa'ideeyaan xogta xaddidan?
    • Iyadoo si fiican loo hagaajinayo moodooyinka is-maamulka sida wav2vec 2.0
    • Adigoo iska ilaalinaya gebi ahaanba shabakadaha neerfaha
    • Adigoo isticmaalaya xeerar gacanta lagu qoray oo keliya
    • Adigoo duubaya malaayiin muuqaallo dareen oo cusub marka hore
  6. What does the EU AI Act prohibit regarding emotion recognition?
    • All emotion-related research in any setting
    • Inferring emotions in workplace or educational settings, except for medical or safety reasons
    • Every voice assistant that detects speech
    • Any system that transcribes spoken words
  7. Why can a speech-emotion model trained on one dataset fail on another?
    • Emotion labels are identical across every corpus
    • Microphone and speaker differences never affect audio
    • Speakers, language, recording conditions, elicitation, and label distributions can differ
    • A trained model cannot process new recordings
  8. What is one limitation of treating a speech-emotion prediction as a person’s true internal state?
    • Acoustic features contain no information
    • A predicted label is a clinical diagnosis by default
    • The same voice always expresses the same emotion
    • A model infers patterns from audio and may not reflect the speaker’s context or self-report