FlowPocket
Tools › Audio tools › Filler Word Analyzer

🗣️Filler Word Analyzer

Transcribes your speech with AI and counts how many times filler words like "um" and "uh" show up — useful for reviewing presentation or interview practice.

Your audio is never sent to a server (the AI model itself is downloaded from a third-party site the first time you use this tool)

Provide audio (choose a file, or record live)

Settings
Words to count (comma-separated, editable)

How to use

  1. Choose an audio file, or record your speech live with "Record with microphone."
  2. Edit the list of filler words to count if needed (the default list works fine as-is).
  3. Press "Analyze" — the AI transcribes the audio, then counts the filler words. The AI model downloads on first use only.

AI transcription plus simple word counting

This tool transcribes your audio with an on-device AI (Whisper) and then counts how many times your specified words (like "um," "uh") appear in the text — it doesn't judge the quality of your speaking, just counts occurrences mechanically. It's useful for objectively reviewing filler-word habits in recorded presentation or interview practice.

"Speaking rate" is a rough guide: transcribed character count divided by audio length in minutes. It's not an absolute measure of "too fast" or "too slow," but it's useful for comparing your own practice sessions over time.

FAQ

Is my audio uploaded anywhere?
No. Transcription and counting all run inside your browser. Only the AI model itself is downloaded from a distribution site on first use — neither your audio nor the result is ever sent anywhere.
Can I change which words are counted?
Yes. Edit the comma-separated list to add or remove whatever words or phrases you like.
What about words like "like" or "so" that have legitimate non-filler uses in English?
This tool counts occurrences without judging context, so ordinary (non-filler) uses get counted too. Because English has more of this ambiguity, the default list sticks to lower-risk fillers like "um" and "uh" only.
Does transcription accuracy affect the results?
Yes. If the AI mishears something, the filler count is affected accordingly. Treat the results as a helpful estimate rather than an exact measurement.

Related tools