TTS Tools
Harness for TTS and voice work: local listening pages to compare clips side by side and save picks
About This Project
TTS-Tools provides a set of local web-based interfaces designed to streamline the evaluation of voice synthesis and audio processing outputs. It allows users to listen to multiple audio clips in parallel, facilitating side-by-side comparisons, ranking, and selection through various templates such as A/B testing, grid-based filtering, and blind-label evaluation. The tool automatically saves user selections as JSON, enabling seamless integration into automated voice design or model training pipelines.
Built using standard Python with zero external dependencies, the project generates self-contained HTML files that can be served locally or opened directly from the disk. It includes specialized templates for tasks such as EQ chain comparisons, model checkpoint evaluation, and terminal term checking. This utility is intended for audio engineers and developers who require a structured, repeatable method for assessing and documenting voice model performance.