Ear prompting looks like a listening skill. It is not. Listening while you speak is odd for about a minute and then it stops being the problem. The problem is distance: staying a fixed number of words behind a recording that has no idea you exist.
Everything actors are taught about ear prompters is a workaround for that one fact. A tape does not wait.
What you are actually working with
An ear prompting system is two parts. Something that plays audio back, and something that puts it in your ear without being visible. The kit varies, the shape does not, and the equipment side is worth its own guide.
What matters for technique is that the playback is fixed. Once you press play, the words arrive at a pace decided minutes or hours ago, by you, in a different room.
Lay the copy down flat
The instinct is to record your script the way you want to perform it, with the emphasis and the pauses already in place. Coaches teach the opposite, and they are right.
Record it evenly. Steady pace, close to monotone, no performance in it at all.
Two reasons. The first is that a performed recording is one you will copy. You will hear your own rising inflection and produce it back, which is how ear prompter delivery gets its distinctive recited quality. A flat track gives you words to say rather than a reading to imitate.
The second is practical. If the client asks for warmth on a particular line, or a beat before the product name, you can give it to them. If you baked your interpretation into the track, you are married to it and the only way to change your mind is to record again.
The count-in
Start the recording with a spoken count. Three, two, one, then the first line.
This is small and it solves a real problem: knowing exactly when to lift your eyes to the lens. Without it you are watching for a cue that arrives without warning, and the first line comes out startled. With it you have three beats to settle, find the camera, and be ready on the word.
Finding your lag
Your lag is how far behind the track you speak. Most people settle somewhere around two or three words. It is personal, it is stable once you find it, and it is the thing that takes practice.
Too close and you are stepping on the recording, which makes you rush and clip the ends of words. Too far and you are holding a queue of words in your head, which is where the flat, disconnected delivery comes from.
Find it on something short. A paragraph you know, read four or five times, paying attention only to the gap. Do not add performance until the gap is automatic.
The four things that go wrong
Echoing. You reproduce the recording's shape instead of speaking. Fix it upstream by laying the copy down flatter.
Drifting ahead. You know what is coming, so you start to anticipate, and now you are racing the track. Slow down and let it lead.
Falling behind, then sprinting. A stumble puts you a sentence back and you speed up to catch the tape. This is the one audiences hear most clearly.
Sounding recited. Everything is technically correct and none of it sounds like a thought. Usually a symptom of a lag that is too long: you are transcribing rather than speaking.
What changes when the track waits
All of the above is the cost of playback that cannot respond to you. The discipline is real and it is worth having, because a hardware rig on a client's set will demand it.
It is not, however, inherent to ear prompting. TellyPrompter's ear prompting mode builds the guide track from your own rehearsal read and then paces it against your delivery. Stall, and it holds. Skip forward, and it seeks to meet you. You set how far ahead it runs — a sentence, or around eight words — which is the same lag decision, made once in a setting rather than maintained continuously in your head.
That takes the distance-keeping off your plate. It does not take away the performance. You still have to sound like the words are arriving now, which is the part nobody can automate, and it still takes a few goes before the sensation feels normal.
Two scope notes. The guide track cannot be captured on Android — it works on iPhone, iPad, Mac and Windows. And voice mode needs the Web Speech API, which Firefox does not have, so use Chrome, Edge or Safari.
If you want to drill the lag before you own any of the kit, practising without the hardware is the cheapest way in. And if you are building this skill because a breakdown asked for it, what casting means by "ear prompter proficient" sets out what is actually being requested.
If you are weighing this against learning the lines outright, the memorise or prompt comparison covers that trade honestly. And if the wooden-delivery problem is what brought you here, most of it is not specific to ear prompting at all — see reading without sounding wooden.