Ikibazo gihuza ibibazo · Biragoye Urwego

VALL-E na Codec Ururimi rwicyitegererezo

Ongera usubiremo VALL-E ya codec-token imvugo-yerekana uburyo, icyerekezo cyayo cyerekanwe, hamwe nibisabwa inyuma yamakuru yatangajwe bike.

Inzira zijyanye no kuyoboraVall E na Codec Ururimi rwicyitegererezo
Ikibazo 1 Bya 8

Ni ikihe gitekerezo cyibanze gitandukanya VALL-E na sisitemu yambere-y-imvugo?

Reba buri kibazo8 ibibazo

Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.

  1. Ni ikihe gitekerezo cyibanze gitandukanya VALL-E na sisitemu yambere-y-imvugo?
    • Ihanura ibimenyetso bya codec neural nkibikorwa byo kwerekana ururimi aho guhanura spekrogramu
    • Ikoresha gusa intoki zerekana ibimenyetso
    • Birasaba amasaha y amajwi avuye kumugambi
    • Itanga inyandiko ivuye mu mvugo
  2. Ni amajwi angahe yerekana VALL-E ikeneye guhuza ijwi ryumuvugizi mushya?
    • Hafi y'isaha imwe
    • Amasegonda atatu
    • Nta majwi yerekana na gato
    • Iminota igera kuri mirongo itatu
  3. Ni ubuhe bwoko bwa codec ya VALL-E ikoresha kugirango ikore amajwi yayo?
    • MP3
    • Opus
    • EnCodec
    • WAV
  4. Kuki VALL-E ikoresha byombi autoregressive hamwe na non-autoregressive stage?
    • Guhindura imvugo mu ndimi nyinshi
    • Gutandukanya umuziki n'imvugo
    • Gucomeka inyandiko
    • Kuberako ibimenyetso bya codec bitondekanye, birahanura ibimenyetso bito bikurikiranye kandi nibimenyetso byiza murwego rumwe
  5. Hafi yamakuru angahe yo kuvuga VALL-E yatojwe, bigafasha gukomera kwa zeru?
    • Amasaha agera ku 60.000
    • Amasaha agera ku 100
    • Amasaha agera ku 1.000
    • Amasaha agera ku 10
  6. Usibye timbre, ni iki kindi VALL-E ishobora kubika muri clip ngufi?
    • Amagambo nyayo avuga
    • Ibidukikije bya acoustic byo gufata amajwi
    • Video yaherekeje
    • Imiterere ya dosiye yumwimerere
  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