Summary of Hypertts: Parameter Efficient Adaptation in Text to Speech Using Hypernetworks, by Yingting Li et al.
HyperTTS: Parameter Efficient Adaptation in Text to Speech using Hypernetworks
by Yingting Li, Rishabh Bhardwaj, Ambuj Mehrish, Bo Cheng, Soujanya Poria
First submitted to arxiv on: 6 Apr 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Machine Learning (cs.LG); Sound (cs.SD); Audio and Speech Processing (eess.AS)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper presents HyperTTS, a neural speech synthesis approach that uses adapters to improve out-of-domain speaker performance in text-to-speech (TTS) tasks. The authors argue that traditional methods of fine-tuning the whole model for each new domain are parameter-inefficient and propose using a hypernetwork to generate parameters of adapter blocks, allowing for dynamic conditioning on speaker representations. The approach is evaluated in two domain adaptation settings, demonstrating state-of-the-art performance in the parameter-efficient regime. The authors also compare different variants of HyperTTS with baselines from previous studies. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary HyperTTS is a new way to make speech synthesis work better when talking to people who sound different from those used during training. Right now, making this kind of system work requires retraining the whole model for each new person’s voice. This can be slow and uses too many computer resources. HyperTTS solves this problem by using a special tool that generates instructions for the speech synthesis model, allowing it to adapt quickly and efficiently to new voices. The results show that HyperTTS performs better than other methods and opens up new possibilities for making speech synthesis systems that work with anyone’s voice. |
Keywords
* Artificial intelligence * Domain adaptation * Fine tuning * Parameter efficient
