Reading, the third trimester
What a phone can and cannot pick up, week by week
Home listening gets easier as pregnancy goes on, for three reasons that have nothing to do with the app. Here are the numbers we measured on public research data, including the band where they got worse, and the place where we stop making claims.
Clip one is a real third trimester recording from the same database these numbers come from.
Firstsong is not a medical device, it is not a doppler and it does not use ultrasound. No week of pregnancy makes it able to tell you that your baby is well. If you have noticed fewer movements than usual, or a change in the pattern you know, call your midwife or maternity unit now, at any stage and at any hour.
What is actually changing as the weeks go by
Three separate things move in your favour across the third trimester, and none of them is software.
- The source gets stronger. A bigger chest with a bigger heart in it puts more energy into the sound, while the room noise, the microphone and the layers in between stay exactly where they were.
- The position settles. Earlier on a baby turns freely, so a spot that worked in the morning is pointing the wrong way by the evening. Later there is less room to turn, so the chest tends to stay where you found it.
- The geometry improves. As the head settles low and the body straightens out along one side, there is a broader stretch of chest lying against the front of your belly instead of a small target that keeps moving.
There is a fourth reason, and it is about us rather than about you. The public evidence that anyone can download and check is overwhelmingly third trimester. That is where we measured, so that is where we are willing to say anything.
The numbers, band by band
We ran the engine over every third trimester abdominal recording in a public research database and split the results by gestational age. This is the whole table, including the row that is worse than the one before it.
| Gestational age | Recordings | Rhythm held | Error against the reference |
|---|---|---|---|
| Below 28 weeks | 4 | not published | no promise is made here |
| 28 to 34 weeks | 20 | 90.0 per cent | 3.52 bpm |
| 34 to 37 weeks | 21 | 85.7 per cent | 2.57 bpm |
| 37 to 42 weeks | 75 | 93.3 per cent | 2.47 bpm |
Look at the middle row. The detection rate goes down before it goes up again, and the honest reading of that is not a story about pregnancy. It is a story about twenty one recordings. Moving a single recording from one column to the other shifts that percentage by about five points, so the dip is noise and anyone who builds a narrative on top of it is selling you something.
Why we say nothing before the third trimester
The same database holds four usable recordings below week 28. Four is not a sample, it is an anecdote with a citation, and it is nowhere near enough to base a promise on.
So we do not make one. At 22 weeks or 24 weeks we do not claim a phone will find anything, we have no measured rate to quote you, and an app that shows you a confident number there is not working from evidence anybody can check. Not everything unproven is false, and phones may well do better earlier than we can currently show. Until we can show it, silence is the only defensible thing to publish.
This is the line most of this category walks straight over. A first trimester heartbeat is the most searched version of this question and the least supportable one, so the pages that answer it confidently are answering it with copywriting rather than with data. We would rather lose that search than earn it dishonestly.
What actually helps, and when
- Wait rather than push. If it is not working now, the same phone in the same room a fortnight later is the single most effective change available to you.
- Follow the kicks. Wherever you feel them, the chest is usually on the opposite end. Kicks up under your ribs generally mean the chest is low.
- Later on, go low. Once the head has settled into the pelvis, the easiest window tends to be below the navel and a little to one side rather than in the middle.
- Quiet room, wired headphones, bare skin, no case. These matter more than the week does, and they cost nothing.
- Check every number against your own pulse. Count for thirty seconds, double it, divide. The full check.
- Stop after ten minutes. The odds do not improve after that and your own pulse rises, which makes a faint signal harder to separate.
When to call your midwife
No week of pregnancy changes this part. What matters is movement, and what matters about movement is your own baby's pattern rather than any count you read online.
Fewer movements than usual, or any change in the pattern you know, is worth a call at any hour, as many times as it happens, and nobody will mind. The NHS guidance is here, and Kicks Count exists for exactly this. A heart rate on a screen, in any week, cannot do what that phone call does.
How Firstsong handles the week you are in
Firstsong asks what week you are in and then changes what it is willing to say. In the third trimester it will publish a rate once enough independent windows agree. Below week 28 it will still listen and still amplify, because hearing your own belly is worth something on its own, but it will not put a fetal number on the screen and it will tell you why in one sentence. That answer is deliberately worse than the competition's, and it is the whole reason to trust the rest of it.
Three public recordings, three verdicts, decided in this tab. Nothing is uploaded.
One email, the day it opens
Firstsong is not on the App Store yet. Leave an address and you get exactly one message when it is, with the price on the first line. No newsletter, no drip, no partner offers. You can reply to that email and a person reads it.
Stored on our own server, never sold, never shared. Your IP is hashed, not kept. Nothing about your pregnancy is asked for or stored.
The bottom line
Across the third trimester a well built engine found a stable rhythm in roughly five recordings out of six, and matched the reference within a couple of beats per minute when it did. That is a genuinely good result, and it was measured with sensors taped to the abdomen rather than with a telephone.
So read the table as the ceiling rather than as a forecast. It tells you the third trimester is the right time to try, that later is easier than earlier, and that anyone quoting you a rate outside this range is quoting something they did not measure.
Common questions
What is the earliest week this works?
We do not have a defensible answer, and that is not modesty. The public database we validated on holds four usable recordings below week 28, which is too few to base anything on. Our numbers start at week 28 because that is where the evidence starts.
Why does the 34 to 37 band look worse?
Because it holds twenty one recordings. One recording moving between columns changes that figure by about five points, so the dip is sampling noise rather than a fact about pregnancy. We publish it anyway, because deleting the inconvenient row is how this category got its reputation.
Does a bigger bump make it easier or harder?
Both, and the two effects push against each other. More tissue between the chest and the microphone costs you signal, while a bigger, stronger heart and a settled position give it back. That is why individual results vary far more than any table can capture.
Are these numbers what I will get on my phone?
No, and this is the limit worth repeating. Those recordings came from dedicated research sensors held against the abdomen, not from an iPhone microphone on real skin. How much survives that change is the measurement we do not have yet, and it will be published here whatever it says.
My friend heard hers much earlier. Was that fake?
Not necessarily. Individual bodies vary enormously, and a thin abdominal wall with a posterior placenta and a well placed baby is a much easier problem than the average. What is not defensible is turning one person's early success into a promise made to everybody.
Keep reading
- What the silence actually means. when the week is right and the phone still finds nothing
- Anterior placenta. the reason two women in the same week get completely different results
- The arm test. how to tell whether an app is measuring anything at all
- Hear the difference. the engine live on a real third trimester recording from this same database
Last checked 2026-08-23