Read the live transcript with confidence-colored text

Read the scrolling speech-to-text transcript in the Copy Assist panel, where each word is color-coded by the whisper.cpp engine's confidence level.

Before you start

• The radio must be connected and an active slice must be receiving audio.
• The Copy Assist panel must be open (View > Copy Assist, or press Ctrl+Shift+T).

Steps

1. In the Copy Assist panel, click Enable / Disable to start the speech-to-text engine on the active slice's audio.
2. Wait for the engine status indicator to show "Listening".
3. Speak into the microphone or receive audio on the active slice.
4. Read the transcript. Each word or phrase is shown in a color that represents the whisper engine's confidence:
5. Green — high confidence
6. Yellow — medium confidence
7. Red — low confidence

What each control does

Control	Behavior
Enable / Disable	Toggle button. Starts or stops the whisper.cpp speech-to-text engine on the active slice's audio. Default: Disabled.
Transcript	Scrolling, read-only text field. Shows decoded speech with confidence-based coloring. The Clear button next to it empties the buffer and also flushes the carried decode context, so the next transcription starts from a clean prompt.
Model tier	Combo box. Selects whisper model size: tiny (default), base, small, or medium. Larger models improve accuracy but use more VRAM/RAM and are slower.
Compute device	Combo box. Selects GPU (CUDA/Metal) or CPU for whisper processing. Default: GPU (CUDA/Metal).
Backlog indicator	Shows seconds of received audio not yet transcribed. Escalates from amber to red as backlog grows.
Settings button	Opens the Copy Assist settings dialog for model tier, compute device, and engine configuration.
Clear	Empties the current transcript buffer and flushes the carried decode context.

Engine status

The engine status indicator shows the current state of the whisper.cpp speech-to-text engine:

State	Meaning
Idle	Engine is stopped or not yet started.
Downloading model	Engine is fetching the selected model file.
Loading model	Engine is loading the model into memory.
Listening	Engine is actively decoding received audio.
Error	Engine encountered a problem and cannot continue.

Tips

• The backlog indicator shows how far behind the engine is. If it stays red for more than a few seconds, a smaller model tier or CPU compute may help on lower-end hardware.
• On first use, the engine may need to download the selected model. Wait for the status indicator to show "Listening" before expecting a transcript.

Related

• Copy Assist — Speech to Text overview
• Enable speech-to-text transcription on a slice
• Change the whisper model tier for accuracy vs speed
• Choose GPU or CPU for speech recognition
• Clear the transcript buffer