Filler Alarm

VERSION 1.0

English · On your device

YOUR SPEAKING SPACE

Ready when you are

MAKE ROOM FOR THE PAUSE

Speak with intention.

Catch hesitation sounds as you speak.

Allow the microphone when prompted. First start downloads about 71 MB.

SESSION TIME00:00

FILLER SOUNDS00

FILLERS / MIN—

Detected sounds

Live practice

Your detections will appear here. The model listens to your voice directly; no transcript is needed.

Check a short audio recording

Choose an English audio clip up to 60 seconds and 20 MB. It is processed locally and is never uploaded.

Audio recording

Make it your practice.

01 / SETUP

Sensitivity

Use Catch more if it misses hesitations. Choose Fewer false alarms if normal speech triggers it.

Sounds to flag

um uh / ah hmm other hesitations

Exact sound names can be uncertain. Words such as “like” and “so” are not counted.

Siren alert

A short siren with each red alert.

Volume25%

Red screen alert

Use headphones to keep the siren out of your microphone.

Microphone

Available microphones appear after you allow access.

Device & performance

Use CPU compatibility mode

Recognition runs on your device. Faster devices give quicker feedback.

Alert delay will appear after a detection.

A practice aid, not a score.

The model can miss fillers or flag normal speech, especially with background noise. Practice in a quiet place and compare its feedback with what you hear.