The AI voice
Voice picking works in Stockly without this extension: the browser’s own speech reads the route out loud, immediately and at no cost. It sounds mechanical, but it reads bin names and numbers clearly.
This extension adds a second voice, listed as AI voice (clip library).
What changes
Section titled “What changes”The voice sounds natural, which matters most in the sentences an operator hears hundreds of times a day and stops parsing word by word.
Each phrase is generated once and kept as a clip. Later waves replay the stored file, so picking never waits for synthesis and a repeated phrase never costs a second call.
Setting it up
Section titled “Setting it up”The voice model sits in the connection wizard, under Voice model (packing assistant): a TTS model, a Voice and a Listen button. → Connections
Without a voice model configured, the packing assistant simply keeps using the browser’s voice. Nothing breaks and nothing is announced — which is worth knowing, because a silent fallback looks exactly like a working setup.
Keep the phrases fixed
Section titled “Keep the phrases fixed”The economics rest on repetition: a phrase said a thousand times is generated once. Feed variable data into a phrase — order numbers, product names — and every announcement is a new phrase, a new call and a new clip that will never be replayed.
The clip library is where that becomes visible. → Voice clips