How do you calculate your speaking speed in words per minute?

Speaking speed in words per minute is the number of words you say divided by the number of minutes you spoke. The arithmetic is easy. The hard part is choosing the minutes. If you time the whole recording, pauses included, you get the overall rate. If you time only the stretches when words are coming out, you get the net rate. In large measured samples of recorded phone conversation, the overall rate was 196 words per minute ¹ and the net rate was 236 ². Both numbers are correct, and they answer different questions.

The steps below give you one number you can repeat and compare. They use a recording and a word count, so you need no special software.

Time a recording, count its words, and divide

1. Record yourself talking for at least two minutes on a topic you know well, such as describing your job. Use the device and the distance you would use in a meeting.
2. Write down the total length of the recording in minutes. Convert any seconds into a fraction of a minute, so that 30 seconds counts as half a minute. This is the input for the overall rate.
3. Make a transcript. A free option is our audio to text tool, which runs in your browser.
4. Count the words in the transcript. Paste it into a word counter, or use the word count in your word processor.
5. Divide the word count by the minutes from step 2. The result is your overall speaking rate in words per minute.

For the net rate, you need the time spent speaking, which is the total time minus the silences. Marking the silences by hand in an audio editor takes longer, but the same division applies once you have that time. Use the same recording for both numbers, so that any difference comes from the pauses.

Pauses decide which number you get

The overall rate and the net rate can differ by a lot, and the difference is pauses. In the conversational telephone speech that researchers Jiahong Yuan, Mark Liberman and Christopher Cieri studied, the overall rate was 196 words per minute, and the net rate with silences and non-speech noise removed was 236 ^{1 2}. Taken together, those two numbers imply that about 17 percent of elapsed conversation time carried no speech ³.

Speech researchers also use articulation rate, which is a different unit. It counts syllables per second and leaves out pauses ⁴, so it cannot be compared directly with words per minute. The glossary entry on articulation rate explains the term.

Count real words, not five-character blocks

The typing convention is different. A typing test scores a word as five characters, spaces and punctuation included ⁵, so a typing score is a count of keys, not of words. Your spoken count is a count of real words in a transcript. A typing score and a spoken count with the same number are therefore not the same measure. The question on how words per minute is calculated explains the typing rule, and the glossary entry on words per minute defines the unit.

The transcript decides what gets counted

Your count depends on the transcript you use. Automatic transcription systems do not handle filler words, repetitions and revisions the same way. A 2025 study of non-native English speech found significant variation across recognition systems in how they handled those disfluencies ⁶. A system may drop "um" and repeated words, or keep them, and the total changes.

Choose a rule before you count, and use it every time:

• Clean verbatim removes filler words and false starts. Use it when you want to count the content you delivered, which is the closer match to a script. The glossary entry on clean verbatim sets out the convention.
• Verbatim keeps every filler word and repetition. Use it when you want the rate of the speech as it was produced. The glossary entry on filler words explains what counts.

Short answers read low, so measure a long stretch

The length of each turn changes the rate. In the same conversational corpus, speaking rate rose rapidly as a speaker's turn grew from one word to seven words ⁷. A recording of short answers, such as "yes" or "right", will therefore read lower than a recording of you explaining something. Aim for at least two minutes of continuous talk, and avoid a test built from one-word replies.

A rough figure of 125 to 150 is not a measurement

You may see a lower figure quoted as normal speaking speed. Jinjuan Feng, Clare-Marie Karat and Andrew Sears give 125 to 150 words per minute ⁸ as people's normal speaking speed, in a dictation study. Treat it as a rough reference, not a measurement. If your count lands between 125 and 196 ¹, that does not mean you are wrong. It means the setting, the task and the counting rule differ from one source to the next.

Which number to report, and how to check it

Report the overall rate when you want to plan time, such as how long a script will run. Report the net rate when you want to know how fast your voice moves while it is moving. Always name which one you used.

A calculator that asks for a word count and a time only divides the two numbers. It cannot tell pauses from speech, so give it the time you spoke if you want the net rate, and the full recording length if you want the overall rate.

Repeat the test in two settings, such as a casual chat and a short talk, and compare the results. The gap between them is often more useful than either number alone. If you are planning a talk, the question on how many words per minute to speak in a presentation explains why a published pace is not a target for every speaker. For the adult reading figures, which are the other half of most speed comparisons, see the question on average reading speed. Our speaking speed guide sets out the full set of measured rates, and the guide to typing speed covers the keyboard side.