Why Everyone Gets Words Per Minute Wrong
A Words Per Minute Chart is just a conversion table that tells you roughly how long a script, transcript, or piece of voiceover copy will take to produce or perform. The most common version you'll see is a simple grid: word count on one side, estimated minutes on the other. It's useful if you're doing transcription work or preparing scripts for clients who want a timing estimate before you commit to recording. Here's the standard version. These numbers come from industry averages used by transcription houses and voiceover agencies: Slow pace (100-110 wpm): suitable for audiobooks, e-learning narration, or content where clarity is the priority. You'll need about 54 to 60 minutes to read 5,000 words at this speed.
Normal pace (130-150 wpm): this is the broadcast standard. News anchors, corporate training videos, and most commercial voiceover work lands here. 5,000 words at 140 wpm comes out to roughly 35 minutes. Fast pace (160-180 wpm): teleprompter reading, YouTube content, or conversational podcasts. 5,000 words at 170 wpm is about 29 minutes, but the speaker will likely need multiple passes to hit those numbers cleanly without running out of breath. Transcription pace (150-200 wpm): this is what you actually hear on recordings. A typical English conversation sits around 150 wpm. A fast-talking podcaster might push 180. Transcribers use the higher end because they're accounting for filler words, overlapping dialogue, and technical jargon that slows everything down.
The catch nobody mentions: none of these numbers hold up when the content isn't conversational. Legal deposition transcripts, medical dictation, and financial earnings calls all run slower because the speaker is reading dense material rather than speaking naturally. I've seen projects where the initial WPM chart estimate was off by nearly 40 percent because the client assumed a 140 wpm baseline for a scripted regulatory document that actually performed closer to 95 wpm. I ended up building a correction factor into my own workflow—multiply the raw word count by 1.35 for formal scripted content before plugging it into the chart.
Get the Full Details
How to Build Your Own
Most people download a generic chart and use it across every project type, which is why their estimates keep being wrong. A proper approach requires three steps and takes about twenty minutes to set up. First, pull five samples from your actual source material. Not similar material, actual material. If you transcribe technical documentation, grab five real documents. If you handle interview recordings, pull five real interviews. This step matters more than anything else because the variance between a sales webinar and a court deposition is massive. Second, time each sample. Don't estimate. Use a stopwatch or a DAW timeline and mark the exact word boundaries. Divide the total word count by the total minutes to get your personal average. Most people find their average differs from the published charts by 15 to 25 percent.
Third, build a chart that accounts for two multipliers. The first is a content-type modifier. Scripted reading runs about 15 percent slower than conversational speech. Second is a fatigue factor. Anything over 45 minutes of continuous recording drops wpm by another 8 to 12 percent because the speaker's articulation slows as they get tired. Apply both to your base average and you'll have a chart that actually predicts real-world output.
Where the Standard Words Per Minute Chart Falls Apart
The biggest pitfall I see people run into repeatedly is applying a single WPM value to mixed-content work. A corporate video that blends scripted narration with unscripted interview segments will never match a consistent word-per-minute rate across its runtime. The narrated portions might sit at 130 wpm while the interview clips drift to 110 because the subject is answering questions rather than performing. Another thing that breaks charts: non-native English speakers or heavily accented presenters. Native-speaker benchmarks assume a certain syllable density and rhythm. When those shift, the wpm drops even if the speaker is perfectly clear. I worked on a project once where the WPM chart estimated a 22-minute audio file and the final deliverable came in at 29 minutes because the presenter was German-accented and the production team refused to re-record. The chart wasn't wrong. The input just didn't match the model. If your work consistently involves heavy scripting, I'd recommend abandoning the WPM chart altogether and switching to a seconds-per-word model instead. That approach measures directly from teleprompter-readable text and tends to stay within 3 percent accuracy, whereas WPM charts drift further as script length increases. For short clips under three minutes the difference is negligible. For anything longer, seconds-per-word pays for itself quickly.
Downloading a Ready-Made Chart
There are several free tables floating around the transcription and voiceover forums. The one I use most often is hosted by the Association for Documentary Editing and it covers five speed tiers with a breakdown for read-aloud versus conversational delivery. You can find it by searching for their editorial productivity guidelines, which are published under a public domain license so there's no paywall or account requirement. If you want something simpler, the transcription industry standard posted on TranscribeMe's resource page has a clean table format that exports to CSV, which makes it easier to drop into a spreadsheet and apply your own multipliers. I keep a copy of that one because it lists both raw word count and adjusted word count for medical and legal content, which covers about 80 percent of the projects I handle. Whatever chart you end up using, remember that it's a starting point, not a contract. The only way to trust your estimates is to log your actual timing data after each project and compare it against the chart prediction. After about ten projects with the same client type, your adjustments become automatic and you'll stop needing the table at all.
