Across
- 6. Neural architecture using self-attention
- 7. RNN variant handling long-term dependencies
- 8. Field that enables machines to mimic human intelligence
- 9. Simplified recurrent neural network
- 12. Converts text into hidden representations
- 14. Mechanism to focus on relevant input parts
- 16. Converts acoustic features into waveform audio
- 17. Frequency-based audio representation for TTS
- 18. Predicts future samples from past outputs
- 20. Computing system inspired by biological neurons
- 23. Network effective in feature extraction
- 25. Converts encoded features into speech output
- 27. Artificial production of human speech
Down
- 1. Raw audio signal representation
- 2. Generating speech from trained model
- 3. Delay between input and speech output
- 4. Process of learning model parameters
- 5. How human-like synthesized speech sounds
- 10. Machine learning using multi-layer neural networks
- 11. Collection of text and audio samples
- 13. Converts written text into natural-sounding speech
- 15. Written representation of a sound
- 19. Intonation, stress, and rhythm in speech
- 21. Network designed for sequence modeling
- 22. Single system trained from text to speech
- 23. Large structured speech dataset
- 24. Basic sound unit of spoken language
- 26. DeepMind model that generates speech one sample at a time