← WOD notes

What a timer voice should sound like: robot vs real cues

Vlad KovnerovFounder, IntertaktOctober 8, 2026Updated October 17, 2026Voice cuesTTSTimer setup

Rest period, round five. The voice starts the call-out for the next exercise, gets as far as the word “next”, and the interval switch eats the rest of the sentence. The count-in you were waiting for never arrives. You start the round on a beep and a guess.

“It cuts off half of the script” (user review of a popular interval timer, Android)

Half. The reviewer had programmed custom call-outs and the app played them, up to every interval switch, where each sentence was cut at the same place: the useful half.

What the voice is for

With the screen off and the phone across the room, everything the timer has to say, it says out loud. Beeps mark transitions. Words carry the information: what comes next, which round. A voice cue is a short sentence, and it has to be finished before the switch it is describing, with a breath to spare for reaction time.

Where the robot comes from

Most timer apps do not choose a voice. They take whatever the phone’s speech engine hands them and move on to the next feature. The default reads “20” the way it reads a street number: flat stress, no air.

Russian-language reviews have a name for the problem: робот-голос, the robot voice. It has been a top complaint about timers there for years. Phone speech in some languages lagged English for a long time, which explains part of it. The rest is simpler. Nobody involved treated the voice as part of the training. The engines have improved every year; the tuning, in most apps, still never happens.

The sentence that does not fit

A cut-off starts as arithmetic. The call-out lives in the window between “rest started” and “work begins”, and it has to fit there with room left for you to react. “3…2…1 rest” fits anywhere. “Get ready for kettlebell swings, starting in three, two, one” fits until the day rest is ten seconds and the exercise name is longer than the exercise. Then the app picks: shorten the sentence, or hold the switch. The bad ones do neither. The voice keeps talking straight through the work it was supposed to introduce.

Anyone who trains to audiobooks knows the price of a sentence from the other side: a bar forgives you and a sentence does not. A voice that respects the clock writes short call-outs and lets numbers do the counting. A long exercise name gets said once.

Three seconds too late

An old German review of a well-known interval timer, from 2018, says the announcement arrives three seconds too late. Picture the sequence. You move on the beep, the effort starts, and then a calm voice tells you to begin, mid-swing. Twice is enough. After that you stop believing the voice, and you are back to squinting at the screen between reps.

Under the music

At a single shared volume, the voice faces a wall of playlist and loses, so the app crams everything louder or forces the music down and keeps it down. Those are the same mistake arriving from two ends: the voice was never given a channel of its own. The fix is old. Give the voice a level above the song, and if the music should dip for a call-out, make the dip brief and make it a setting. What a loud room does to cues and why forced ducking earns one-star reviews each have their own writeup.

What a good voice does

The last bullet reads like the smallest of the five. Train one hard session in a language you only half know and it stops looking small.

Where Intertakt stands

The cues come in English, Russian and Spanish. The voice counts you in and out of every interval and says work and rest out loud. The round is told to you, so you are not the one counting it. Call-outs are written short on purpose, so a phrase finishes ahead of its switch instead of racing it. The voice sits above your music at a level of its own. With ducking on, the dip covers the call-out, the song comes straight back, and there is no lag between the voice and the clock. It keeps talking with the screen locked and DND on.

Intertakt is out on Google Play. If you want one email when the iOS version lands, leave your address on the waitlist.

FAQ

Why does my interval timer sound like a robot?

Most apps use the phone's built-in speech engine and never touch the defaults: the pace and the way numbers are read. On Android you can sometimes improve things yourself, because the system text-to-speech settings let you swap the voice or the whole engine, and a timer that uses the system voice inherits the change. If the app ships its own audio, the voice came from a decision someone on the team made.

Why does my timer cut off its own announcements?

Usually the script outgrew the window. A long exercise name or a chatty call-out cannot fit inside a ten-second rest, and an app that plays it anyway has decided that hearing the first half is better than hearing nothing, which is not true. On your side, shorter custom names and a few seconds more rest fix most of it. On the maker's side, the real work is editing: a cue should be written against the tightest transition in the whole plan.

Can the voice be louder than my music without ducking it?

Yes, when voice and music are separate channels. Set the voice above the track once and it stays there; beeps can sit lower, and music lands wherever you like it. If you do want the song to dip, keep the dip short enough that the beat returns before the next interval, and keep the whole behavior behind a switch you control.

Do interval timer voices work in languages other than English?

Less often than they should. English is the default and other languages are treated as a bonus, so Russian reviews keep running into the same robot voice reading their rounds. The voice either barely exists in the language or pronounces it like a stranger. Intertakt speaks its cues in English, Russian and Spanish, with the same timing in each.

Download

Take the timer to the gym

Intertakt is free on Google Play: unlimited sessions, no ads, screen off.

Download on Google Play