The best apps for learning French, by what you're actually doing

The gap that makes French different

Here's a sentence a second-year learner reads without blinking: Il y a des amis qui nous attendent.

Now here's roughly what hits your ear at speed: yadezami kinouzatond.

That's ordinary French, spoken normally, at the speed a colleague would use. Three separate processes conspired to erase the word boundaries your eyes rely on.

Liaison revives a silent final consonant when the next word starts with a vowel. Nous avons becomes nouzavon, the Z arriving from a letter you were told not to pronounce. Elision deletes a vowel and fuses two words: je ai was never an option, it's j'ai. Enchaînement takes a consonant that is pronounced and hands it to the next word's first syllable, so the break you expected lands in the wrong place.

And then ordinary speech compresses on top of all that. Il y a collapses toward ya. Je ne sais pas comes out closer to chais pas. Both are normal. Neither is what you studied.

Spanish barely does this to you. Spanish spelling maps onto Spanish sound with unusual honesty, so a Spanish learner's reading ability and listening ability climb together. In French they separate early and the gap widens for years.

This has been measured rather than merely observed. Annie Tremblay and Elsa Spinelli tracked English speakers' eye movements while they listened to French, and found the acoustic cue distinguishing a liaison consonant is genuinely there and genuinely faint: a liaison consonant runs about 15% shorter than the same consonant at the start of a word. On an ordinary consonant that difference is a matter of milliseconds. The learners in their study leaned on that duration cue harder than the native speakers did, which Tremblay and Spinelli put down to English already using duration to find word edges.

So the cue exists, it's tiny, and your ear reaches for it because it's the tool English gave you. No amount of vocabulary study sharpens it. Keep that in mind while reading every recommendation below.

The uncomfortable answer first

Two concessions, before I recommend anything, because the rest of the page is worth less without them.

The first: if you're a beginner without a tutor, Verbamor is the wrong purchase. It's built to catch what a tutor taught you and it has nothing to catch before then. Skip to the beginner section and buy something else.

The second matters more, and it applies to everyone reading. Nothing on this list fixes listening comprehension as fast as large amounts of real native audio. No flashcard app substitutes for that, mine included. If your problem is parsing spoken French, a subscription to any of these is a smaller lever than daily listening to French media plus a tutor who corrects your ear. I'd bet an hour of French radio a day for 3 months does more for your segmentation than any deck ever built.

Cards do a different job: they hold the individual words still while your ear learns the shape of the stream. I'd rather tell you that than sell you the wrong ordering.

Beginner, no tutor: Pimsleur, or Babbel for rules

Pimsleur is a stronger first recommendation for French than it would be for most languages, and the reason is structural rather than a matter of taste.

It's audio-only. You never see the spelling before you've heard the word, which sidesteps the specific trap French sets: read beaucoup or oiseau or eaux first and your English reading habits will build a pronunciation that has almost nothing to do with the real one. Bassetti and Atkinson found that the written form keeps bending pronunciation even in learners with years of instruction behind them, so this isn't a beginner-only tax. It's a tax you avoid by hearing things first, which I've argued at length in sound before spelling.

The method has real spaced-repetition ancestry too, and it deserves conceding. Four named principles: anticipation, graduated interval recall, core vocabulary, organic learning. Graduated interval recall means learners "are reminded of new words at gradually increasing intervals," which is what a modern scheduler automates. Pimsleur was running it on cassette long before any of these apps existed.

French runs to 5 levels, with reading lessons and conversational lessons on top, and the first week is free. On price, Pimsleur's own subscriptions page publishes no dollar figure; its blog says both consumer tiers cost "around $20 per month." That's the only first-party number available, and it's deliberately approximate. Any page quoting you $19.95 to the cent got it from a third-party blog.

The limits are real. The content is fixed, and the pace is slow by design.

Babbel is the pick for a different learner: the adult who wants the rule stated before being asked to use it. French rewards that more than Spanish does, because liaison and elision have teachable conditions. Knowing that liaison is obligatory between a plural determiner and its noun, as in les amis, and forbidden after et, converts some of the noise into structure.

Babbel says most of its lessons take around 10 to 15 minutes, it ships voice recognition and pronunciation guidance, and its Review feature does use spaced repetition, pulling vocabulary from lessons you've finished and scheduling sessions for you. Babbel doesn't publish its interval ladder, so I can't tell you how it compares to anything, and no page claiming to has actually seen it.

Price: 1, 3, 6 and 12-month plans plus a lifetime option. The annual works out around $9 a month, though that figure comes off a Babbel coupon page, so treat it as a discounted rate rather than a standing one. Babbel discounts more or less permanently. And if you found Babbel while shopping for a teacher, note that Babbel Live is no longer available to individual consumers.

You freeze on real speech: Lingopie, with a caveat

A specific and common profile: you've done a course app, you can read, and the first time you put on a French film without subtitles you understood roughly one word in ten and felt the floor go.

Lingopie is built for that gap. It's French television with clickable subtitles, and every word you click gets saved into flashcards and quizzes. It carries French among 15 languages, with playback-speed control, which for French is the feature that matters most: dropping a scene to 80% speed re-inserts the word boundaries that full speed erased.

Now the caveat, and it's the reason this section has one. Subtitles let you read French while believing you're hearing it. Your eyes are fast and they'll happily do all the work, at which point you finish an episode feeling comprehension you don't have, and the next unsubtitled conversation puts you straight back on the floor.

The fix is a discipline the product won't impose on you: watch each scene once with no subtitles, then again with them, then a third time with none. The first pass is the one doing the work. The middle pass is the answer key, and Karpicke and Roediger's results below are about exactly this distinction between producing something and being shown it.

Nobody corrects your ear: italki

Every app on this page speaks to you at a speed chosen to be understood. A French person will not.

That's the case for a tutor, and it's sharper in French than in Spanish. You need someone who talks at unmodified speed, notices the exact moment your face goes blank, and can tell you which of the 3 processes above just ate the sentence. Nothing automated does that. A tutor also catches the errors that are invisible from the inside: the liaison you're inserting where French forbids it, the final consonant you're pronouncing that should be silent.

italki needs no subscription and has no bulk minimum. You buy credits and book. Tutors set their own prices across two tiers, Community Tutors (native or fluent, no formal credentials, conversation-focused) and Professional Teachers (formally qualified). Here's where that number comes from, since it's the one price on this page I can't take off a pricing table. italki's own French tutor page says French tutors "cost between $4 to $60 per hour," which is a band wide enough to be useless. So I read the live listings on 11 August 2026 and wrote down the trial prices in front of me: $5.00, $7.00, $8.00, $12.50, $15.00, $15.00, $15.00, $20.00, $23.49, $25.00, $29.00. Trials run cheaper than standing rates, community tutors sit at the bottom, and a full hour from a professional teacher realistically lands around $15 to $30. Check it yourself; the listings are public and they'll have moved by the time you read this. The optional italki Plus ($5.99 a month, or $4.99 billed annually) adds AI study tools, and you don't need it to book anything.

Preply is the other big marketplace, and the difference is the billing shape rather than the quality. Per Preply's own help centre, a subscription is tied to one tutor and renews automatically in 28-day cycles with a minimum of 4 lessons per cycle, and lessons you don't schedule before the renewal date expire without carrying over. That commitment suits a fixed weekly habit and is genuinely harder to quietly abandon. If your weeks are irregular, or you want to hear 4 different French accents before committing to one, italki's credits are the safer shape.

I won't publish Preply hourly rates. Tutors set their own, the figures circulating come from third-party roundups rather than Preply, and an invented number is worse than an absent one.

Reading fine, hearing badly: Anki with audio

You read Le Monde with a dictionary open. You follow a slow podcast. Then a French colleague says something at speed and you catch the shape and lose the content, and more Babbel doesn't move you.

Course apps have a finite syllabus and yours is finished with you. What you need next is words chosen from the specific things you couldn't catch, which no fixed course can know about.

Anki ships no French content at all. That emptiness is exactly what an intermediate needs. It's a scheduler you fill from your own listening, and building the cards is itself doing work: Slamecka and Graf named the generation effect in 1978, and material you generate is remembered better than the same material read.

The French-specific instruction is this: put audio on the front of every card. A French card with only text on the front is testing your eyes, and your eyes are the part that already works. Make the prompt a native recording of the whole phrase, so ils ont and ils sont stop being a spelling difference and become the sound difference they actually are. Learn phrases rather than bare words, too, because liaison lives in the seam between words and a word in isolation hides it entirely.

Here's the correction I'd most want an intermediate learner to have, because a hundred SEO posts get it backwards. Anki's default scheduler is SM-2, Wozniak's 1987 SuperMemo algorithm. FSRS has been built into Anki since version 23.10 in October 2023 and it's very good, but the manual calls it "an alternative to Anki's legacy SuperMemo 2 (SM-2) algorithm" and tells you to enable it at the bottom of the deck options page. Install Anki, never open the settings, and you're running a formula from 1987. Turn it on and you get FSRS-6: 21 parameters fitted to your own review history, with a desired retention target you set, defaulting to 90%. The mechanics are in the FSRS post.

Cost: free on Windows, macOS, Linux and Android with free AnkiWeb sync. AnkiMobile on iOS is US$24.99 one time, and regional prices differ.

Two things to read before you start filling it. French gives English speakers an enormous vocabulary head start through shared Latin and Norman ancestry, and knowing which words you already own for free changes what belongs in a deck (cognates and false friends). And decks are half an intermediate plan at most, because at this level listening volume does work no card does, which I've argued out in input versus cards.

Already paying for lessons: Anki, or Verbamor

This is the one segment where I'd shortlist my own app, and it still isn't automatic.

The problem is specific. You pay $25 for an hour, your tutor corrects 6 things, you understand all 6 in the room while she's saying them, and by Tuesday 4 of the 6 are gone. The hour did its job. The week undid most of it while you weren't looking.

The cleanest evidence on what stops that slide comes from a vocabulary experiment. Karpicke and Roediger gave students 40 Swahili-English pairs and then changed only one thing: what happened after a pair was first answered correctly. Pairs that kept getting tested returned at about 80% a week later. Pairs that were merely restudied returned at about 36%.

Karpicke & Roediger 2008 · vocabulary recall, one week later

0%

recalled when items kept being tested after the first correct answer

0%

recalled when items were only restudied after the first correct answer

That last detail is the uncomfortable one. They had no felt sense of which method was working, which is how "I'll just listen to the recording again" survives so many weeks of not working.

Timing has numbers behind it too. Cepeda and colleagues, across more than 1,350 people, found the best-performing gap sat at roughly 20 to 40% of a 1-week test delay, falling to about 5 to 10% at a year. On a weekly lesson cycle that points at reviewing a day or two after you walk out, rather than the night before you go back. A week sits at the very short end of what they measured, so read it as a direction rather than a prescription.

Anki does this job for free and does it well, and where it falls over is arithmetic. A good French card with real audio on the front takes a few minutes to build, it gets built after the lesson when you're tired, and 2 skipped weeks make a backlog that ends the habit. If you're the person who assembles 9 audio cards every week for a year, you don't need me and I'd rather say so.

Verbamor is for the other person. It records the lesson, mines the words and phrases you actually got wrong, builds cards from those, and schedules them with FSRS so each returns near the day it would have gone. Every card carries native audio of its example sentence, and the recall side plays the sentence with the answer replaced by a tone, so you're reconstructing a French word from the sound of the phrase around it. For a language where the difficulty lives in the seam between words, that's the format I'd defend.

Now the disclosure, because this is where comparison pages usually start lying. Anki publishes a scheduler anyone can inspect: FSRS-6, 21 parameters, 90% default retention. Verbamor runs FSRS as well, and that is the end of what I can prove. I have no retention figure of my own to show you. Too few people have used it for long enough to produce a number that would survive scrutiny, and a percentage invented to fill a table is worth less than an empty cell. Everything measured on this page was measured by Tremblay, Karpicke, Cepeda, Bassetti and Slamecka, and it's evidence that retrieval, spacing and ear-first learning work, not proof that my app runs them well.

Having published everyone else's prices, here are mine. Explorer is $19.99 a month. Native is $129.99 a year, about $10.83 a month, with a 7-day free trial on the annual plan. There's no free tier and it's iOS only. Both of those are real reasons to buy something else.

The longer version of that segment is in language apps for tutored learners. If you're weighing a free course app against lessons, that trade is in Verbamor vs Duolingo. And the equivalent page for Spanish sorts differently, because Spanish learners have a genuinely different problem: the best apps to learn Spanish.

All 6, side by side

Why there's no score out of 5 here

The page currently ranking first for this query hands each app a single blended number: 4.8, 4.6, 4.5. No rubric breakdown, no hours logged per app, no test dates, no screenshots. A number with no components can't be checked, which is convenient for whoever wrote it.

It does have a named author with an M.Ed. and 25 years of teaching experience, which is more than most of this SERP can say, and it's why it ranks. The credential is in teaching Spanish though, on a page about French, and the page never once mentions liaison or listening comprehension. That's the gap this page exists to fill.

I didn't run a study, so I'm not printing one. What I can do is show the work. Every number on this page is traceable: to a first-party page the company publishes, or to a paper with a DOI you can open. Where a number wasn't verifiable I left the hole open and said why.

Which is why you'll notice deliberate gaps. No Pimsleur price to the cent. No Babbel interval ladder, because Babbel hasn't published one. No Lingopie price, because its homepage doesn't carry one. Every number in a competitor's table on those points was invented somewhere, and you can verify that in about 10 minutes.

Questions people actually ask

---

Sources

Every load-bearing claim Verbamor makes is traced to its paper on the research page. Prices and product facts on this page were checked against first-party sources on 11 August 2026.