Type or say a meal without searching a food list#
On Today there is always a box that says Log anything…. You can type what you ate as a sentence, "two eggs on toast, half an avocado and a flat white", or tap the microphone and say it. The app turns that into an estimate with each food as its own item, weight and all, and logs it. There is no food list to search, no brand to pick from twenty near-duplicates, and no recipe to rebuild ingredient by ingredient first.
The reason is not that search is always slow. For a packaged snack with a barcode, a matching database entry can make logging straightforward. The reason is that the meals people rated hardest to record, homemade dishes, restaurant plates and food at other people's houses, are ones where respondents described difficulty finding matching entries, and describing a meal in your own words is how you would explain it to anyone else. The case for a light record at all is in the pillar on mindful eating with numbers.
Which meals are hard to log with a database#
A study of 141 current and lapsed food journalers asked people to rate how easy different kinds of meals were to record, on a 7-point scale where higher meant easier1. The ranking is the argument:
| Kind of meal | Ease rating (1–7, higher is easier) |
|---|---|
| Packaged food | 6.5 |
| Fast food | 6.3 |
| Home-cooked meals | 4.6 |
| Buffet meals | 3.7 |
| Ethnic food | 3.7 |
| Restaurant meals | 3.6 |
| Foods served by friends | 3.2 |
| Foods at parties | 2.9 |
Packaged and fast food scored significantly easier than everything else. The authors called this a negative nudge: some respondents reported choosing easier-to-log foods over foods they considered healthier.
The comments explain why. Home cooking meant rebuilding a recipe: "If a particular item isn't in my app's database, I would have to deconstruct it into its constituent ingredients and then add those individually," and "recipe builders took a long time." Large databases created their own friction, too many near-matches and not the plain food someone actually ate: "Oh, was it McDonald's oatmeal? Quaker Instant apple cinnamon oatmeal in a pouch? No. It was normal oatmeal!" Restaurant food added a knowledge gap: "Meals at a restaurant are very difficult because you do not know the ingredients."
A sentence sidesteps most of this. "Normal oatmeal with milk and a banana" needs no brand. "Chicken stir-fry, about a plate, rice on the side" needs no recipe builder. The estimate will be less certain than a scanned label, and it says so with a range, but the meal gets recorded. Estimating a home-cooked plate well is its own skill, covered in estimating homemade meal calories.
A food database can make the granola bar easy and the family dinner hard. Both can be entered as a sentence.
How much time logging takes, and what the format changes#
Logging is a daily time cost. In a 24-week online weight-loss program with 142 participants, average time on the self-monitoring page was 23.2 minutes a day in month one; among the 93 participants still monitoring in month six, it was 14.6 minutes a day2. That time included food, exercise and weight entries. Time spent monitoring fell over the program; the study could not establish why, whether from efficiency or from recording fewer foods or less detail. It did not break the time down by step, so it cannot say how much was searching, how much was typing, or how much was reviewing entries already made. That leaves the question of which step costs most open for any single logging method.
In a crossover study, 18 young women kept three 7-day food records, on a computer, on a smartphone and on paper, in random order, and each record was compared with estimated total energy expenditure, derived from measured resting expenditure and accelerometer-based activity3. All three under-reported by similar amounts, roughly 456 to 510 kcal a day, reaching 78% to 80% of expenditure. The study found no statistically significant accuracy difference between the three formats. Preference differed: 89% liked paper least. The sample is small and young, so treat it as a signal. Participants preferred the digital formats; whether that preference improves long-term logging was not established.
Neither study tested sentence-style logging. Put together, they say the time cost is real and that under-reporting appeared whichever format was used. The trade between fast and detailed entries is weighed in quick-add vs detailed logging.
What typing a meal looks like#
- One sentence, several foods. "Greek yogurt with honey and walnuts, then a coffee with milk" comes back as separate items, each with its weight, calories and protein. Phrases like "X with Y" stay as the two foods they describe.
- Quantities in your terms. Write "two eggs", "a handful of nuts" or "150 g chicken"; stated amounts are used as given, and amounts the app had to assume are drawn dashed so you can see and change them.
- Nothing to finish first. The meal logs as soon as the estimate appears, with an Undo on the receipt. Why the app gives you the number first and asks nothing is explained in get a number, then correct it.
What saying a meal looks like#
- Live captions. Tap the microphone and your words appear as you speak, so you can see a misheard word before you finish.
- Start over, safely. If the first sentence went wrong, the round ↺ Start over button clears the transcript, and a toast offers Undo for five seconds in case you change your mind.
- Pause, edit, resume. You can pause, tap into the transcript to fix a word by typing, and resume dictating from the end.
- Your own vocabulary. Local dish names, family shorthand and rough amounts are all fair input; you do not need to translate "grandma's cabbage soup" into a product name before logging it. If a word is misheard, fix it in the transcript first.
- Any mix of languages. The recognition setting is Auto · mixed, so a sentence that switches language halfway is still one meal.
These studies do not establish whether spoken logs are less accurate than typed ones, so check the transcript and review the foods and amounts before relying on the estimate. The wider evidence on spoken food records is in voice logging and food accuracy.
FAQ#
Can I log food by typing a normal sentence?#
Yes. Type the meal the way you would say it, including several foods and rough amounts, and it comes back as separate editable items with calories and protein.
Why are homemade meals so hard to log in most apps?#
Respondents in one survey described difficulty finding matching entries and entering individual ingredients, and rated home-cooked meals harder to log than packaged food.
Can I fix a misheard word before the meal is logged?#
Yes. Pause dictation, tap into the transcript and type the correction, then resume or finish. Review the foods and amounts on the receipt before relying on the estimate.
What if I say the wrong thing while dictating?#
Tap Start over to clear the transcript, or pause and edit the text directly. If you clear it by mistake, you have five seconds to tap Undo and restore the transcript.
Sources#
- Cordeiro F, Epstein DA, Thomaz E, et al. Barriers and negative nudges: exploring challenges in food journaling. Proc SIGCHI Conf Hum Factor Comput Syst. 2015;2015:1159-1162.
- Harvey J, Krukowski R, Priest J, West D. Log often, lose more: electronic dietary self-monitoring for weight loss. Obesity (Silver Spring). 2019;27(3):380-384.
- Hutchesson MJ, Rollo ME, Callister R, Collins CE. Self-monitoring of dietary intake by young women: online food records completed on computer or smartphone are as accurate as paper-based food records but more acceptable. J Acad Nutr Diet. 2015;115(1):87-94.
Source: BurnWeek — "Typing or saying a meal in BurnWeek", https://burnweek.fit/blog/logging-food-in-your-own-words/. Licensed CC BY 4.0: free to quote or reuse with a link to this page.
