Tortoise TTS Autoregressive Synthesis Quiz

Answers and explanations appear as you take the quiz.

1. What does Tortoise’s autoregressive transformer do in the synthesis pipeline?
   • Predicts speech tokens conditioned on text and speaker information
   • Measures the final waveform’s loudness only
   • Selects a license for the generated audio
   • Converts video frames into subtitles
2. What can Tortoise do with just a few short reference clips?
   • Transcribe audio to text
   • Clone the target voice and read arbitrary text
   • Translate languages only
   • Remove background noise
3. Which two model families most influenced Tortoise's design?
   • Decision trees and SVMs
   • Rule-based phoneme tables and HMMs
   • Autoregressive transformers and diffusion
   • Convolutional filters only
4. How does Tortoise choose among multiple generated candidates?
   • It picks the first one
   • It always picks the shortest
   • It asks the user
   • It uses a CLVP scoring model to rank them against the text
5. Who created Tortoise TTS?
   • James Betker
   • Geoffrey Hinton
   • Ian Goodfellow
   • Andrej Karpathy
6. What is the main reason Tortoise is relatively slow?
   • It downloads voices from the cloud each time
   • It prioritizes quality, sampling and refining many candidates
   • It runs only on CPUs
   • It compresses audio repeatedly
7. What can Tortoise’s random voice option generate without a supplied reference speaker?
   • A verified recording of a specific real person
   • A translation of an existing speaker’s legal identity
   • A synthetic voice selected by the model
   • A speaker embedding from a missing audio file
8. How does Tortoise package reference-voice conditioning for reuse?
   • It stores only a text transcript
   • It saves the voice as a LoRA for an image model
   • It records the speaker’s legal name in the waveform
   • It stores autoregressive and diffusion conditioning latents