Kedis ku xidhan hagaha · Adag Heerka

VALL-E iyo Qaababka Luuqadaha Codka Kediska

Dib u eeg qaabka qaabka luuqada VALL-E's codec-token, jiilkeeda calaamada, iyo xaaladaha ka dambeeya natiijooyinka cod yar ee la soo sheegay.

Wadooyinka haga ee la xidhiidhaNoocyada Luuqadaha Vall E iyo Codec
Su'aal 1 ee 8

Waa maxay fikradda asaasiga ah ee VALL-E ka soocaya nidaamyada hore ee qoraalka-ilaa-hadalka?

Horudhac su'aal kasta8 su'aalo

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

  1. Waa maxay fikradda asaasiga ah ee VALL-E ka soocaya nidaamyada hore ee qoraalka-ilaa-hadalka?
    • Waxay saadaalisaa calaamadaha codec neural sida hawsha qaabaynta luqadda halkii ay saadaalin lahayd spectrogram
    • Waxay adeegsataa hab-samaynta calaamadaha gacanta oo keliya
    • Waxay u baahan tahay saacado cod ah oo laga helo afhayeenka bartilmaameedka ah
    • Waxay ka soo saartaa qoraalka hadalka
  2. Intee in le'eg ayay cod-bixineed VALL-E u baahan tahay si ay u xidho codka afhayeenka cusub?
    • Qiyaastii hal saac
    • Ilaa saddex ilbiriqsi
    • Ma jiro cod tixraac ah haba yaraatee
    • Qiyaastii soddon daqiiqo
  3. Qoddobka neural codec ee VALL-E u adeegsato soo saarista calaamooyinka maqalka?
    • MP3
    • Opus
    • EnCodec
    • WAV
  4. Waa maxay sababta VALL-E ay u isticmaasho marxaladda autoregressive iyo mid aan toos ahayn?
    • Si hadalka loogu turjumo luqado badan
    • Si loo kala saaro muusiga iyo hadalka
    • Si loo cadaadiyo gelinta qoraalka
    • Sababtoo ah calaamadaha codec waa lakabyo, sidaas darteed waxay saadaalisaa calaamado aan dhab ahayn si isku xigta iyo calaamado fiican oo isbarbar socda
  5. Qiyaastii intee in le'eg xogta hadalka ayaa VALL-E lagu tababaray, taasoo awood u siinaysa cloning eber?
    • Qiyaastii 60,000 saacadood
    • Qiyaastii 100 saacadood
    • Qiyaastii 1,000 saacadood
    • Qiyaastii 10 saacadood
  6. Timbre ka sokoow, maxaa kale oo VALL-E ka ilaalin kartaa goos gooska gaaban ee tixraaca?
    • Erayada saxda ah ee afhayeenka weligii
    • Deegaanka dhawaaqa ee duubista
    • Muuqaalkii la socday
    • Qaabka faylka asalka ah
  7. What does VALL-E condition on to imitate a target speaker in its demonstration setup?
    • A photograph and a written biography only
    • A long labeled transcript from every target-speaker recording
    • A short audio prompt from that speaker plus the text to speak
    • A speaker ID with no audio example
  8. What text representation does the VALL-E paper use as input to its text-to-code model?
    • A raw waveform of the target utterance
    • A speaker’s face embedding
    • A list of MIDI notes
    • A phoneme sequence