Summary of Anonymising Elderly and Pathological Speech: Voice Conversion Using Ddsp and Query-by-example, by Suhita Ghosh et al.
Anonymising Elderly and Pathological Speech: Voice Conversion Using DDSP and Query-by-Example
by Suhita Ghosh, Melanie Jouaiti, Arnab Das, Yamini Sinha, Tim Polzehl, Ingo Siegert, Sebastian Stober
First submitted to arxiv on: 20 Oct 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: Sound (cs.SD); Audio and Speech Processing (eess.AS); Quantitative Methods (q-bio.QM)
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 This paper proposes a novel approach to speech anonymization that can effectively retain linguistic content, prosody, and unique speech patterns found in elderly and pathological speech domains. The method, called DDSP-QbE (differentiable digital signal processing with query-by-example), uses differentiable signal processing and query-by-example training with novel losses to disentangle linguistic, prosodic, and domain representations. This enables the model to adapt to uncommon speech patterns and outperforms state-of-the-art voice conversion methods in terms of intelligibility, prosody, and domain preservation across diverse datasets, pathologies, and speakers. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps protect speaker identity by changing personal identifiers in speech while keeping the language understandable. The current ways of doing this don’t work well for older people or those with health problems. To fix this, the researchers created a new method that uses voice conversion and special training to keep prosody and unique speech patterns. This approach was tested on different datasets, pathologies, and speakers, and it showed significant improvements in understanding, prosody, and domain preservation while maintaining quality and speaker anonymity. |
Keywords
» Artificial intelligence » Signal processing
