Start with a phrase, review your positive and negative sample counts, then launch the training pipeline with a live console so every step is visible.
Name the wake phrase and choose the language/voice set used to generate training audio.
Device-captured positives and reviewed negatives are used when present. Manual samples are managed from the Samples tab.
Listen to clips sent by your sats, turn good wake-word examples into personal samples, and save false positives as negatives for the next training run.
Refresh clips from the trainer inbox and keep an eye on how many reviewed negatives and personal samples are ready.
Approve strong wake-word clips into personal samples, or mark false positives as reviewed negatives.
Listen to saved positive and negative samples, remove anything that does not belong, or manually import seed recordings when you need a starting point.
Personal samples are positive examples. Negative samples are false wakes and hard negatives.
Optional backup path for seeding positives before your device has captured enough real wake audio.
WAV, MP3, M4A, FLAC, OGG, AAC, OPUS, and WEBM are all fine when ffmpeg is available. Files are converted into personal_samples/.
personal_samples/.Flash only for firmware updates. If this is a new satellite or it is not already on Tater firmware 3.0.3 or newer, do one USB flash first, then use this tab for fast OTA updates.
No reflash is needed after training. Copy a trained wake word JSON URL below and paste it into your satellite's Home Assistant microWakeWord Model URL entity.
Choose a prebuilt Tater firmware image from the shared firmware repo.
Auto-detect a device or enter the OTA target manually.
The console shows download, verification, OTA upload, USB flash, and reboot progress.
http://127.0.0.1:8789 in Chrome or Edge; port 3232 is only for OTA updates after firmware is installed.