Kedis ku xidhan hagaha · Adag Heerka

Kediska Jiilka Muusikada Kala Sarreysa ee MusicLM

Dib u eeg calaamadaha kala sareynta ee MusicLM, qaboojinta laxanka, toosinta qoraalka-maqalka MuLan, iyo agabka qiimaynta MusicCaps.

Wadooyinka haga ee la xidhiidhaJiilka Muusigga Heerarka ee Musiclm
Su'aal 1 ee 8

How does MusicLM formulate conditional music generation?

Horudhac su'aal kasta8 su'aalo

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

  1. How does MusicLM formulate conditional music generation?
    • As hierarchical sequence-to-sequence modeling over audio tokens
    • As painting pixels with a diffusion model
    • As concatenating fixed prerecorded loops
    • As simulating instrument mechanics
  2. Waa maxay doorka calaamadaha semantic ee MusicLM?
    • Waxay kaydiyaan xogta badan ee faylka heesta
    • Waxay qabtaan qaab dhismeed heer sare ah sida laxanka iyo laxanka
    • Waxay u tarjumaan erayada luuqado kale
    • Waxay xakameeyaan mugga wax soo saarka
  3. Qaybtee ayaa bixisa tafaasiisha dhawaaqa wanaagsan sida timbre iyo texture?
    • Qoraalku si toos ah ayuu u degdegay
    • Tallaabada dib u dhiska ee Griffin-Lim
    • Codka dhawaaqa neural codec ee SoundStream
    • Buug sharci aragti muusik oo gacanta lagu qoray
  4. Sidee MusicLM ugu xidhaa qoraalka degdega ah maqalka muusiga?
    • By ereyada muhiimka ah oo u dhigma cinwaannada heesaha
    • Adiga oo waydiisanaya bini'aadam inuu calaamadeeyo clip kasta
    • Iyada oo loo marayo API tarjumaad gaar ah
    • Iyada oo loo marayo MuLan, meel dhex galka muusig-qoraalka wadajirka ah
  5. Waa maxay sababta kala sarraynta, naqshadeynta marxaladda ay ka caawiso qaab-dhismeedka fog-dheer?
    • Waxay u ogolaataa calaamado aan dhab ahayn, oo gaabis ah inay qaabeeyaan qaab wakhti dheer ah iyada oo aan lahayn dherer isku xigxig oo weyn
    • Waxay meesha ka saaraysaa baahida loo qabo xog kasta oo tababar ah
    • Si dhakhso leh ayey u ciyaareysaa heesta si ay ugu habboonaato in badan
    • Kaliya waxay dhalinaysaa saddex ilbiriqsi markiiba
  6. What does the MusicLM paper report about its generated music duration and sample rate?
    • Every output is exactly 30 seconds at 8 kHz
    • Its experiments report 24 kHz music that can remain consistent over several minutes
    • Only one-second clips at 48 kHz
    • Uncompressed MIDI of unlimited duration
  7. What additional input can MusicLM use alongside text to guide generation?
    • A spectrogram-only image caption
    • A microphone-array steering vector
    • A hummed or whistled melody
    • A source-code repository
  8. What did the MusicLM paper release as MusicCaps?
    • A collection of 5,500 synthesizer presets
    • A speech transcription benchmark
    • A set of SoundStream model weights only
    • 5,500 music-text pairs with human-written descriptions