ParrotPal Protein calculator

Voice logging calories for 13 days to make weight: a case study

I had 13 days to get from 84.2 kg to under 82.3 kg for a jiu-jitsu tournament, with two weeks of meals out and socialising in the way. I logged everything I ate by voice in ParrotPal and weighed myself every morning. I weighed in at 81.4 kg, with 0.9 kg to spare.

First published 28 April 2024 · Updated 7 October 2026

The result

March–April 2024
Start84.2 kg25 March
Limit82.3 kg6 April
Weigh-in81.4 kg6 April
Under the limit0.9 kg13 days logged

One person’s numbers, logged by voice only and weighed every morning. Most of a quick drop like this is water and food in the gut, not fat.

Why I ran this test

The question we were asked most in 2024 was simple: is logging food by voice accurate enough to rely on? Typing every food into a database is slow, so most people give up on it. Saying “a large bowl of pasta with chicken and a glass of red” takes seconds. But if the numbers that come back are wrong, quick is useless.

So I gave it a test with stakes. I had entered a Brazilian jiu-jitsu tournament with a weigh-in limit of 82.3 kg on Saturday 6 April 2024. Miss it and I’d be disqualified and lose my entry fee, with friends and family travelling to watch. On Monday 25 March I weighed 84.2 kg.

The rules I set myself:

The numbers, day by day

My March log showed a clear pattern. Weekends spiked: on Saturday 23 March, two days before the test began, I logged about 4,100 kcal. Weekdays came back down to a little below maintenance, so my weekly average landed a little below maintenance.

What the log and the scale showed
Date (2024)Logged by voiceScale
Sat 23 MarchAbout 4,100 kcal, a typical big weekend day–
Mon 25 MarchThe test starts: every meal logged by voice from here84.2 kg
Tue 2 AprilLogging every meal, social plans includedStill over 84 kg
The last four daysTwo days under 2,000 kcal, then about 2,200 kcal on each of the two days before the weigh-inA sharp drop
Sat 6 AprilWeigh-in, then a burger and chips to celebrate: close to 4,000 kcal81.4 kg

Figures from my ParrotPal log and daily weigh-ins, as first published on this page in April 2024.

Four days out I was still over 84 kg, which wasn’t comfortable. But the log showed me exactly where the extra calories were coming from, so I knew what to change: a few lower days, nothing drastic. The scale followed.

To be clear about what happened: this wasn’t 2.8 kg of fat. A drop that quick is mostly less water and less food in the gut. My calorie average across those weeks was only a little below maintenance. What the test shows is that the logged numbers moved with the scale, day after day.

What it showed

The log matched the scale. When my logged intake went up, my weight followed, and when it came down, so did the scale. That only happens if the numbers going in are close enough to the truth.

Eating out didn’t break it. Most of my social meals had no calories on the menu. Describing them in plain words was enough to keep the trend useful.

Consistency beat precision. I logged every single day because it took seconds. A rough log you keep beats a perfect one you abandon by Wednesday.

The bigger lesson was the weekends. Without the log I’d have guessed I ate “pretty well”. With it, I could see that my weekdays were fine and my weekends weren’t, and that a couple of lower days would bring my week back in line. Awareness changes behaviour, but only if you trust the data and can be bothered to collect it.

Oh, and I came second.

How accurate is voice logging?

Like every calorie tracker, voice logging gives an estimate. Food labels, restaurant menus and database entries are all estimates too. The 2024 version of this page quoted our internal accuracy tests as a percentage; we’ve taken that out, because one number can’t cover every food and every way of describing it.

What matters more is whether the estimate is close enough, often enough, for your weekly average to be right. In this test it was: the logged calories tracked the scale well enough to make a hard weight limit, while eating out. That’s one person over 13 days, not a study, so treat it as a real-world example rather than proof.

You get better numbers by saying what you know. If you know the weight, say it. If you don’t, plain portions like “a handful”, “a small bowl” or “half a plate” are fine, and naming the restaurant or brand helps.

Why easy logging works: ability beats motivation

Stanford behaviour scientist BJ Fogg sums up why we do things in one line: a behaviour happens when motivation, ability and a prompt come together at the same moment. When something is very easy, you’ll do it even when your motivation is low. When it’s hard, you need a lot of motivation to bother.

Think of your phone buzzing on the bedside table at night. It’s right there, picking it up takes no effort, and you want to see who it is, so you always do. Now imagine the phone is downstairs. Most nights you wouldn’t get up.

Traditional food logging is the phone downstairs: weighing, searching and picking the right entry from a list takes real effort, and motivation fades after the first week. Saying what you ate takes about as much effort as sending a voice note to a friend. That’s why I managed it every day for 13 days, through a busy social fortnight.

What’s changed since 2024

ParrotPal has grown up since this test. You can now log food by talking to it, snapping a photo, typing a quick message or scanning a barcode. It’s built around protein first, because when you’re eating less, protein and strength work help you keep your muscle while the fat comes off. Your LeanShield™ score, from 1 to 100, rates the habits linked to protecting muscle: eating enough protein, doing strength work and eating enough overall. And it suggests workouts to lift that score.

If I ran this test again today, I’d still log by voice. I’d also make sure I hit my protein target on the lower days, so more of what came off was water and fat rather than muscle. Work out yours with the protein calculator. If you train jiu-jitsu, here’s how many calories BJJ burns, and if you’re weighing up apps, how ParrotPal compares as a MyFitnessPal alternative.

Try voice logging yourself

Say what you ate and ParrotPal works out the calories and protein. 7-day free trial on iPhone, then a subscription.

Try ParrotPal free Work out my protein Get your free LeanShield score

The ParrotPal app: 4.7★ on the App Store · 700,000+ people · iPhone only.

Questions people ask

Can you log calories by voice?

Yes. In ParrotPal you say what you ate, like “a large bowl of porridge with a banana” or “half a pizza and a pint”, and it works out the calories and protein. Portions in plain English, like a handful or a large bowl, are fine. Like every calorie tracker, it gives an estimate.

Is voice logging accurate enough to lose weight?

In this one-person case study, the logged calories followed the scale closely enough to make an 82.3 kg weigh-in with 0.9 kg to spare, while eating out a lot. It is one person, not a study. What matters most is logging consistently, so your weekly average is close to the truth.

Was the weight lost in this case study fat?

Mostly not. Losing 2.8 kg in under two weeks is mainly water and food in the gut. Fat loss comes from a weekly calorie average a little below what you burn, over months.

Is ParrotPal on Android?

No. ParrotPal is only available on iPhone.

Also from ParrotPal

Please read. This page is one person’s experience and general information, not medical advice or advice on making weight. Quick weight cuts, especially by cutting fluids, carry real risks: plan any cut for competition with your coach, and a GP or sports dietitian if you have a health condition. If you have diabetes, kidney disease, a history of disordered eating or any other medical condition, check with your GP before changing how you eat. Calorie figures are estimates.

MyFitnessPal is a trademark of its owner. ParrotPal isn’t affiliated with or endorsed by it. LeanShield™ is ParrotPal’s muscle protection score: it rates the habits linked to protecting muscle, is not a medical device and doesn’t measure muscle.