Transformer Lab shows local TTS fine-tuning

This is less about a flashy model and more about product maturity: Transformer Lab is making multimodal training feel like a normal app workflow instead of a notebook-only ritual.

• –The local-provider flow is the headline feature for teams that want training to stay on their own hardware or inside their own network
• –Showing the full loop from data prep to audio sampling is the right abstraction for research tooling; it reduces friction at the exact points where training demos usually fall apart
• –GUI plus agent-friendly CLI is a strong combo: onboard humans with the interface, automate repeatable runs with the command line
• –Using Orpheus 3B, a real dataset, and an eval dataset makes the demo feel practical rather than synthetic
• –The open-source positioning matters here because TTS experimentation is still expensive enough that reproducibility and local control are real differentiators