1. Training Pipeline Configuration
"Train Both" runs sequentially.
W: 0 P: 0
0.0001
Total Records for Training
--
Audio Lengths
--
Format Health
--
Corrupt Rows
--
Record Types (A: Sentence, L: Lexicon)
--
2
8

Note: Whisper defaults to 2 (recommended 2ā€“4 to prevent 8GB RTX 4060 Ti VRAM overflow/OOM). Piper is lightweight and defaults safely to 8.

2. Live Training Terminal Output
IDLE
TRAINING PROGRESS: 0.0% (Step 0 / 0)
CURRENT LOSS: --.---- ESTIMATED TIME: --:--:--
[SYSTEM READY] Awaiting training initiation... Select training target, verify export manifest paths, and click "Start Training".
3. Laboratory Sanity Check