Reimagining service delivery in the agentic era with Google Public Sector
cloud.google.com·19h ago
AI Research
arXiv CS.AI·Jun 10·5 min read
Interpreting and Steering a Text-to-Speech Language Model with Sparse Autoencoders
TL;DR
Text-to-speech (TTS) systems lack interpretability in how they generate speech from text. Sparse autoencoders (SAE) were developed to analyze and control features in the CosyVoice3 language model.
✦ Why It Matters
Engineers can use sparse autoencoders to gain interpretability and control over text-to-speech systems.
Key Takeaways
- Utilize sparse autoencoders for better feature interpretation in TTS.
- Explore causal manipulation of speech attributes in your applications.
- Increase laughter probability significantly using targeted interventions.
- Control speaker gender perception while preserving spoken content.
How It Works
The study employs BatchTopK sparse autoencoders to analyze the language model's latent space, allowing for the identification of features that can be manipulated. By labeling features based on their activation context, the model can distinguish between text and speech influences, enabling targeted adjustments to speech characteristics.