How do you speak more clearly on calls?
Why colleagues hear "fifty" when you said "fifteen": a quick test to find your weak spot, drills for mumbling, breath and pace, a five-minute warm-up, and when the problem is your mic chain instead.
You say "we shipped fifteen fixes this sprint" and the product manager writes down fifty. Nobody notices until the retro. On a call the two words differ by one stressed syllable and a final "n" that your laptop mic, the noise filter and the other person's earbuds all treat as optional. In a room the listener would have read your lips and filled the gap. On Zoom they get a compressed signal, a delay, and no second chance unless they interrupt you.
So "how to speak clearly" on a call is really two questions. One is about your mouth: breath, jaw, tongue, pace, how far your voice carries. The other is about the chain your voice travels through before it reaches anyone. Most advice online covers only the first. This guide covers both, starts with the one you control for free, and tells you how to find out which one is actually failing.
If people never ask you to repeat yourself but say your voice sounds thin, boomy or muffled, you want How do you make your voice sound better on calls and recordings? instead. That page is about tone. This one is about being understood the first time.
First, find out which kind of unclear you are
"Speak more clearly" is useless advice until you know what the listener is actually missing. Record yourself the way a colleague hears you: Zoom's Settings > Audio > Test microphone records a few seconds and plays them back, Microsoft Teams has Make a test call under Settings > Devices, and Google Meet's green room only shows a level bar, so for Meet record a voice memo instead. How do you test your mic and hear what others hear on a call? walks through every app.
Read one ordinary work sentence with a number and a name in it, play it back, and match what you hear to this table.
| What people say to you | What is usually happening | Where the fix lives |
|---|---|---|
| "Sorry, can you repeat that?" on numbers and names | Consonants are soft, the stressed syllable is flat | Diction, below |
| "You cut out at the end" | Your volume falls on the last words and a noise filter or gate treats them as silence | Breath, plus filter settings |
| "Slow down a bit" | You are above about 165 words a minute | Pace |
| "You're a bit quiet" | Low projection, or the input level is low | Projection, then mic level |
| "You sound far away" | Mic distance and room echo, not your speech | Muffled mic |
| Nobody complains, but you sound flat | Monotone delivery | Intonation and warm-ups |
The "quiet" and "far away" rows can save you weeks of tongue twisters aimed at a problem that lives in the hardware and the room. If the recording sounds like you are in a bathroom, no diction drill will fix it.
How to stop mumbling: the three mechanical causes
Mumbling is not a personality trait. It is three physical habits, and each one has a drill.
The jaw barely opens. Vowels are shaped by the space inside your mouth. Talk through a half-closed jaw and "ship" and "shop", "cap" and "cop" drift toward the same neutral sound. Laptop users make it worse by looking down at the screen, chin toward the chest, talking into the keyboard. Raise the screen or the camera to eye level before anything else; it costs nothing and puts your mouth back in front of the mic.
The tongue and lips do half the work. Consonants are where the information is on a call. The difference between "fifteen" and "fifty" is a stress shift plus a closing "n"; between "asked" and "ask" it is a "t" the size of a click. The classic drill is the pen exercise: hold a pen or a cork lightly across your back teeth, read a paragraph aloud for thirty seconds so the tongue and lips have to fight for every consonant, then take it out and read the same paragraph again. The second read will probably feel exaggerated. Record both reads and compare them: the gap between how it feels from the inside and how it plays back is the point of the drill.
The energy dies before the sentence does. Many people start a sentence at a good level and finish it in a murmur, because the breath ran out or because the important part felt said already. In a room that fade is fine. On a call it collides with software: Zoom's background noise suppression and the noise gates in OBS and similar tools are built to remove whatever they judge to be background, and a trailing murmur can land on the wrong side of that judgment. OBS's own Noise Gate filter, for example, ships with a close threshold of -32 dB, and when your level falls below it the output is silence, not quieter speech. A word you say under that line is a word that does not leave your Mac. That is the mechanism behind "you cut out at the end", and why the next section is about breath, not volume.
Breathing: fuel for the last word
The US National Institute on Deafness and Other Communication Disorders (NIDCD) puts it plainly in its voice care guidance: support your voice with deep breaths and do not rely on your throat alone. The throat-only version is what makes a voice go thin and pinched by the end of a long standup.
Three habits change this without any singing-teacher vocabulary:
- Breathe before the sentence, not in the middle of a word. Glance at your next point, inhale through the nose while you look, then speak. The pause reads as composure on a call, not hesitation.
- Shorten the phrase, not the breath. If you regularly run out of air before a full stop, your sentences are too long for speech. Split "we reviewed the logs and found the issue was in the cache layer which we fixed on Tuesday" into two sentences. Each one now ends with air left.
- Practise a long, even exhale. Breathe in for a count of four, then let the air out on a steady "sss" and count how long it stays even. Do it three times. The number matters less than the steadiness; a hiss that wobbles or fades early is the same pattern as a sentence that fades early.
The NIDCD page also lists the dull basics that make every exercise easier: drink water, keep indoor air around 30 percent humidity, rest your voice between long calls, and avoid the extremes of shouting and whispering, both of which strain it.
Pace: the one number worth measuring
Microsoft's Speaker Coach, the rehearsal tool built into PowerPoint, recommends that presenters speak at 100 to 165 words per minute, and notes that some people speak comfortably and clearly at other rates. That range turns pace into something you can check with a stopwatch.
Take a paragraph of exactly 150 words from something you actually present, an update, a pitch, a lesson. Read it aloud at your natural call speed and time it.
150 words / 165 wpm = 55 seconds (upper edge of the range)
150 words / 100 wpm = 90 seconds (lower edge)
Under 55 s -> you are rushing; listeners on a lagging call fall behind
55 to 90 s -> fine; work on pauses, not speed
Over 90 s -> you are dragging, or pausing mid-phrase instead of between phrases
Speed alone is rarely the whole problem. Calls add a delay, so a listener who wants to ask something waits for a gap. If you never leave one, they either interrupt you or give up. Leave a beat after every number, name, date and decision. Say the number, stop, then move on. If it matters, say it twice in two forms: "fifteen, one-five". Pilots and support agents do the same with spelling alphabets for exactly this reason: "B as in Bravo" survives a bad line, "B" alone does not.
How to project your voice without shouting
"How to project your voice" usually gets answered for a stage: aim at the back row, fill the hall. On a call the microphone is a few inches from your face, so projection means something narrower. It means a voice that arrives at the same level from the first word to the last, pointed forward, with enough energy that nobody needs to turn you up.
Shouting gets the opposite. A forced voice sounds strained, it clips cheap mics, and the NIDCD lists talking above noise as a direct cause of vocal strain. There is also a reflex working against you: the Lombard effect, the involuntary tendency of speakers to raise their vocal effort in loud surroundings. In a noisy open office you will push without noticing, which is exactly the talking-over-noise strain the NIDCD warns about. Lowering the noise around you, or moving the mic closer so you can speak at normal effort, beats trying to override the reflex.
What projection looks like in practice:
- Mic at a fixed distance. A common cause of "your volume keeps going up and down" is leaning in and out. A headset or a clip-on mic keeps the distance constant for you.
- Speak to a person, not to the screen. Pick the face of one listener and talk to them. A steady level is easier to hold for one listener than for a grid of tiles.
- Watch the auto-level. Zoom has an "Automatically adjust microphone volume" option and its own help page suggests turning it off if your audio fades in and out. If your sentences already fade at the end, auto-level only blurs the picture; switch it off while you practise so the test recording shows what your voice is doing.
If you want the deeper, steadier, broadcast version of a projected voice, mic technique and processing take over from here. How do you get a radio voice on a microphone? covers distance, compression and why radio voices sound the way they do.
Vocal warm ups for speaking: five minutes before a call
Singers warm up and speakers mostly do not, then wonder why the first two minutes of a meeting come out croaky. A speaking warm-up does not need a piano. The useful part is a family of exercises voice specialists call semi-occluded vocal tract (SOVT) work: humming, lip trills, buzzing through a straw. According to VoiceScienceWorks, partly closing the mouth sends back pressure to the vocal folds, which helps them vibrate with less effort, and a narrower straw creates more of that pressure.
0:00 Lip trill (lips loose, "brrr"), slide up and down your range 60 s
1:00 Hum through a drinking straw on "ooo", easy pitch slides 60 s
2:00 Hum "mmm" and open into "ma, me, mi, mo, mu" 45 s
2:45 Tongue twisters, slow first, then at speaking speed 60 s
"Red leather, yellow leather"
"Unique New York"
"Six thick thistle sticks"
3:45 Read the first two sentences of what you will say, twice 45 s
4:30 Water, then join the call
The last step is the one people skip and the one that pays most. Rehearsing your opening line means the first thing the room hears is the version you already said twice.
Diction exercises that survive a real call
Tongue twisters train agility, but a call punishes specific sounds. Practise the ones that carry meaning in your work:
- Word endings. "Next steps", "asked", "texts", "costs", "fixed". Read a list of past-tense verbs and plurals and make every final consonant audible without adding an extra vowel.
- Numbers. Teens against tens: thirteen and thirty, fifteen and fifty, eighteen and eighty. Stress the "-teen" clearly, and when the number decides something, repeat it digit by digit.
- Names. People's names and product names are the words a listener cannot guess from context. Slow down on them and spell the unusual ones.
- Over-articulate, then back off. Read a paragraph with exaggerated consonants, almost theatrical, then read it at normal speed while keeping about half of that effort. This is the same trick as the pen drill: you recalibrate what "normal" feels like from the inside.
Ten minutes a day in the second week of the plan below, with recordings at the start, middle and end to compare: the recordings are the only honest judge, because your own voice in your head comes through bone as well as air and sounds different from what a mic picks up.
How to improve speaking voice when the problem is the chain
Here is the uncomfortable part. You can speak with perfect diction and still lose the listener's attention, because people judge the sound of a voice separately from its words. In a 2025 study in PNAS by Walter-Terrill and colleagues at Yale, listeners heard the same statement either clear or distorted the way a poor microphone distorts it. They transcribed both versions almost equally well, 98.07 percent against 98.09 percent of words correct, yet the clear version was rated more hireable: 76 against 68.5 out of 100. Being understood and sounding good are two different scores.
So after the drills, look at the path your voice travels:
- Bluetooth headsets, AirPods included, are limited in how much detail they can send while their mic is in use. If your test recording sounds dull through a Bluetooth headset, plug in a wired one and record the same sentence. That is the cheapest comparison you can make.
- Aggressive noise suppression can bite into soft consonants. Zoom's community forum has a thread of users reporting a lisp-like effect they blame on it; it is a user report, not Zoom documentation, but if you hear it in your test recording, try a lower suppression level.
- Gates set too high cut sentence endings, as described above. If you use OBS, What are the right noise gate settings in OBS? has the numbers.
This is where TunedMic fits, and also where it stops. TunedMic is a Mac app that adds a virtual microphone tuned to your voice: you read a short passage for about a minute during Sound Check, it takes more than 30 measurements of your voice, mic and room, and every app that picks the TunedMic mic hears the processed version. It has a clarity control and nine profiles, three of them in a Narrator family built for calls and lessons. It cannot make you stop swallowing word endings. Processing works on the sound you give it.
Voice training app or a phone recorder?
People searching for a "voice training app" usually want feedback they cannot give themselves. There are three levels of it.
| Tool | What it gives you | Cost and limits |
|---|---|---|
| Phone voice memo or Zoom's Test microphone | The raw truth: how you sound to others | Free; you judge yourself, so you need a checklist |
| PowerPoint Speaker Coach | Live feedback on pace, filler words and monotone, then a Rehearsal Report with statistics | Free preview on the web, iOS and Android with a Microsoft account sign-in, also in the Microsoft 365 desktop apps; English only; needs internet, and your speech is sent to Microsoft; the report disappears when you close it, so screenshot it; built for slide presentations |
| TunedMic Voice Training | Short exercises for breathing, diction and intonation; record a take and play it back with a checklist of what to listen for | Part of the $19 one-time app; Mac with Apple silicon and macOS 14 or later only, no Windows |
The phone recorder is enough if you are disciplined. Speaker Coach is the best free option if you rehearse presentations in English, since it measures pace against that 100 to 165 range for you. TunedMic's Voice Training makes sense if you also want the call itself to sound better, because the same app handles both: you practise in Voice Training, and the processed voice goes out to Zoom, Meet, Teams, Slack or Discord through the TunedMic microphone. The license is $19 once, covers two Macs and a year of updates. There is a free seven-day trial with every feature included and no card or account, and payments are final, so the trial is the time to decide.
A two-week plan
| Days | Daily routine | Time |
|---|---|---|
| 1 | Record a 60-second update, run the 150-word timing test, fill in the diagnosis table | 10 min |
| 2 to 7 | Five-minute warm-up before your first call; pen drill once a day | 7 min |
| 8 | Record the same update again and compare with day 1 | 5 min |
| 9 to 13 | Warm-up plus the diction list that matches your weak spot (endings, numbers or names) | 10 min |
| 14 | Third recording; if the words are clear but the sound is thin or noisy, move to the mic chain | 5 min |
Start with the recording, not the exercises. Tomorrow, before your first call, open Zoom's Test microphone or a voice memo and read your opening sentence with one number and one name in it. If the number is ambiguous on playback, the drills above are your fix. If the words are clear and the sound is not, watch the 38-second video on the TunedMic home page, which shows Sound Check and the Speech Report you get after a call, then judge on your own mic during the free 7-day trial whether the tuned sound is what you are missing.