How accurate is a word-count duration estimate?

It is a close approximation. Real duration varies with punctuation, pauses, emphasis, and the specific voice, but a words-per-minute model lands within a few seconds for most natural-pace narration.

What speaking rate should I use?

Conversational narration averages about 150 words per minute. Ads and energetic reads run 160 to 180, audiobooks and calm explainers sit around 130 to 140, and IVR prompts are slower still for clarity.

Why estimate credit cost from characters?

Most AI voice platforms bill per character generated, not per word. The tool estimates characters from your word count and multiplies by typical per-character rates so you can budget before generating.

Is my script uploaded anywhere?

No. You only enter a word count, and all math runs in your browser. No script text is sent or stored.

AI Audio Duration Estimator

AI audio duration estimator

Before you spend AI voice credits, it helps to know roughly how long your audio will run and what it will cost. This estimator turns a word count into an estimated duration for different content types and speaking rates, then approximates the character-based credit cost charged by common AI voice tools.

How it works

Spoken duration is driven by speaking rate, measured in words per minute (WPM). A relaxed audiobook read averages around 135 WPM, conversational narration about 150, and a punchy ad read 170 or more. The tool divides your word count by the selected WPM to get minutes, then formats minutes and seconds. For cost, it estimates characters (English averages roughly 5.8 characters per word including spaces) and multiplies by a per-character rate that approximates each platform’s billing tier, so you get a budget figure in credits.

Notes and examples

A 500-word podcast segment at 150 WPM runs about 3 minutes 20 seconds.
The same 500 words as an audiobook at 135 WPM stretches to about 3 minutes 42 seconds — slower pace, longer file.
IVR and phone prompts should be estimated at a slow rate; callers need time to process menu options, so the extra seconds are intentional.
Treat the credit figure as a planning estimate. Pauses, SSML breaks, and re-generations all add to real usage.