MorphFader: Enabling Fine-Grained Semantic Control for Text-to-Audio Morphing through Fader-like Interactions

Supplementary Webpage For

MorphFader: Enabling Fine-Grained Semantic Control for
Text-to-Audio Morphing through Fader-like Interactions

(Webpage best viewed on Chrome or Firefox. For screens smaller than 1200px please use Firefox only.)

Demo Video

Note: All links below have been anonymised.

Index

1. Qualitative Results

1.1 Semantic Word Weighting (opens in a new tab/window)
1.2 Audio Morphing (opens in a new tab/window)
1.3 Comparison with Baselines
- 1.3.1 Comparing smoothness or linearity of morphing (opens in a new tab/window)
- 1.3.2 Comparing trivial prompt editing (opens in a new tab/window)

2. Interactive Interfaces

Proof-of-concept interfaces for the proposed interaction algorithms.
Kindly note that these interfaces are deployed on a single shared RTX 3090 GPU running in a lab environment. Hence, the response time may vary based on the number of users accessing the interfaces and the GPU simultaneously. We recommend using the Google Colaboratory notebooks (section below) for a seamless experience.

3. Interaction Algorithm Notebooks

4. Codebase
5. Perceptual Listening Tests