Shadowing apps compared: which ones let language learners actually shadow, and how to shadow with any podcast
By James Corrigan, English Teacher · Sep 2, 2026 · 12 min read
What a shadowing app has to do
A shadowing app for language learners needs five things: a human model voice at natural speed, a transcript you can hide, pitch-preserving speed control, a one-sentence loop, and a recorder that plays your attempt beside the original. Of the eight dedicated apps I checked in September 2026, only Speak Pro lists all five; most of the rest are English-only.
I teach English and set shadowing as homework. I have watched students buy an app, shadow a synthetic voice for a fortnight, and come back sounding the same; a nurse of mine had shadowed one clip forty times without once hearing herself, and her first playback did more than the forty passes. The app was rarely the problem. The missing recorder was. The method is Sarah Mitchell’s, in The Shadowing Technique: A Complete Guide; this piece judges the tools.
Lambert’s 1992 definition is the specification: paced auditory tracking, speaking what you hear while it is still playing. Five features make that possible on a phone. One name recurs: Yo Hamada, whose 2016 study is the one most shadowing teachers cite.
• A human model voice at natural speed: synthetic will do for a beginner, and our own free tool uses your device’s voices, but the goal is a native recording with the swallowed syllables a speech engine tidies away.
• A synced transcript you can hide: Hamada’s rule is that learners should generally shadow without the script, because reading splits attention between letters, sounds and meaning. I let students keep it until they can keep up; that allowance is mine, not his.
• Speed control that keeps the pitch: any modern player does this. The useful range is 0.75x to 1.0x; at 0.5x the rhythm you are copying no longer exists.
• A sentence loop or A-B repeat: you will need one clause eight times, and scrubbing back with a thumb is how sessions die.
• Record and compare: he also recommends an IC recorder or a phone voice-memo app so students can hear which words they managed to shadow. Playing both back is the feedback; a score is optional, and the research section explains why.
Dedicated shadowing apps compared
Everything below comes from the developers’ own listings, checked in September 2026. Where a listing does not mention a feature I say so rather than claiming it is absent. Tier names only; the figures change too often to print.
• Speak Pro (iOS; not the Speak lessons app from Best AI Apps to Practice Speaking English (2026), as far as either page says): English in American and British voices, Spanish, French, German, Italian, Japanese and Korean, with Dutch on a separate card. Paste a YouTube link, get a synced transcript with tap-to-translate, slow down, loop, record and have your take aligned against the original, then AI feedback. Hiding the transcript is not mentioned; nor is a price, nor Android.
• ShadowingEnglish (web, iOS, Android): English as the target language. Paste a YouTube video, a library sorted by CEFR level, a pronunciation score per repeated sentence, a separate AI tutor conversation mode and a leaderboard. Transcript sync, speed control, looping and price are not described.
• shadowing.app (web, English): upload your own audio, smart pauses, side-by-side comparison, paid for in credits that feedback and uploads consume. A free tier of 10 credits, then three paid tiers. Speed control and looping are not mentioned.
• Tubeshad (iOS, Android): import any YouTube video, repeat sentence by sentence, speech-to-text pronunciation feedback, translation and dictionary. Its Google Play listing is titled Shadowing English; the homepage states neither languages nor pricing.
• Speechling (iOS, Android, web): all six languages Lucida teaches, plus four more. Listen-and-record drills with side-by-side playback and a human coach who corrects your recordings. The free plan includes the curriculum, the comparison and some monthly coaching, except for English, French, Chinese and German; Unlimited is the paid plan. More repeat-after-me than shadowing, but the human ear is real.
• LingoSon (mobile web): English, German, French, Spanish, Turkish, Italian and Portuguese. It generates a dialogue from a typed topic and reads it in a synthetic voice with word highlighting at 0.25x to 1.5x. No recorder, no speech feedback. Three free lessons a month, then two paid tiers.
• Shadowing Player: Languages (iOS): imports audio and video files and YouTube, creates and edits SRT subtitles, loops a section, adjusts speed in 0.01 steps, and has a listen-and-repeat pause mode. Recording and hiding subtitles are not mentioned. Ad removal is a one-off purchase; Korean-first interface.
The verdict from the checklist
The eighth, Shadowing - English Speaking Exercise by Picup (iOS, English and Japanese, paid after a one-month trial), earns a line only because its App Store listing showed version 1.0.4 dated July 2017 when I checked. Confirm the update date before paying.
Score the list against the five mechanics and the field thins to one. Speak Pro is the only listing that describes all five, and I am counting its synced transcript as item two although nothing says you can hide it. Shadowing Player manages four, with no recorder mentioned. Two apps cover all six languages Lucida teaches: LingoSon cannot record you, which removes the one check the research asks for; Speechling can, and plays your take beside the native one.
Free web tools that get you most of the way
Lucida’s own page, Shadowing Practice: Listen, Repeat and Record, Free, needs no sign-up: English passages from A2 to B2, Spanish at A2 and B1, a voice you can slow down, a chunk mode, a loop and a recorder that keeps the audio on your device. The voice is synthetic, so it fails the first item, and it passes three of the other four: there is no button to hide the text, so look away once you can keep up. I am biased about it, obviously.
ShadowingEnglish’s free practice pages run YouTube clips sentence by sentence with a score each. YouGlish finds captioned YouTube clips of any phrase you search, in more than 25 languages, free. Language Reactor is a Chrome extension for Netflix and YouTube: dual subtitles, a pop-up dictionary, replay the current line, auto-pause after each; the core is free, Pro adds saved phrases and more of its 27 languages.
Of those, only our page and ShadowingEnglish capture your voice, and ShadowingEnglish returns a score rather than side-by-side playback, as far as its site describes. The rest give you the model, the text and the loop; you bring Voice Memos on a second phone, not a mental note that you sounded fine.
How to shadow with any podcast, no shadowing app required
The controls you are paying an app for already sit in your podcast player. Apple Podcasts has generated time-synced transcripts since 2024, and as of September 2026 its list covers all six languages Lucida teaches: each word lights up as it plays, and tapping a line jumps to it. Speed is a preset menu or a press-and-drag control. Spotify has a speed button for podcasts and skips 15 seconds either way; third-party guides say 0.5x to 3.5x, but Spotify’s own help page does not list the range, so I have not verified it.
YouTube on the web has presets from 0.25x to 2x and, since December 2024, a Custom slider in 0.05x steps, plus captions and a transcript panel. That covers mechanics one to four in all six languages, provided the episode has text. Mechanic five is a second device, which is Hamada’s own suggestion: Voice Memos on an iPhone or Recorder on Android.
The session itself is in Sarah’s guide; the only change is that the loop and speed buttons come from the podcast app. Five or six passes of one passage is the ceiling before a plateau, a 2010 finding by Shiki and colleagues that Hamada’s guide repeats; then change the text. Six passes of a 45-second clip is four and a half minutes of speaking. In my classes, two short sessions a week for a month is where I hear the first change: eight sessions, well under two hours, for a podcast you already follow.
Which tool for which language
English learners can skip this section; every tool here works for them. For the other five languages the picture is thinner, and the search results make it worse: shadowing.app, ShadowingEnglish, Tubeshad and Picup dominate the results for the query ‘shadowing app’ and are English-only or English-first, so a German learner typing it is mostly shown products that cannot serve them.
Spanish, French and German: Speak Pro, Speechling, LingoSon, Language Reactor, YouGlish, and Apple Podcasts transcripts for the podcast route. Italian: the same six. Portuguese: Speechling, LingoSon, Language Reactor, YouGlish and Apple Podcasts; Speak Pro does not list it. The podcast route is the equaliser, the one workflow identical in all six.
What the research says shadowing improves, and what an AI score is measuring
The best summary is a 2025 systematic review by Whitworth and Rose, with three verdicts: fluency improves; prosody, meaning intonation and rhythm, mostly improves; individual sounds are unproven. Accent is mixed: Foote and McDonough (2017) had 16 advanced ESL learners shadow short dialogues on iPods for eight weeks, recording themselves; imitation, comprehensibility and fluency improved, accentedness did not, while the review reports accent gains elsewhere.
That fits the target the CEFR Companion Volume sets for phonological control: intelligibility, not a native accent. At B2 your accent still shows your other languages but has little or no effect on being understood. So a per-word pronunciation percentage looks odd. It grades the segmental layer, where the evidence is weakest, and it grades a copy of someone else’s sentence.
The gains the evidence supports, rhythm and fluency, are what you hear when you play your recording next to the original. I would rather have playback with no number than a number with no playback. The nurse’s dropped word endings were audible in one playback. No percentage had told her.
The one thing no shadowing app does
The product test I apply to every tool above: does it ever ask you a question you have to answer in your own words? None does. Each plays you a sentence and grades your copy of it. ShadowingEnglish’s AI tutor mode is the exception, and it is not shadowing.
That is the definition of the method, not a flaw in the apps. Shadowing trains delivery: the rhythm, the linking, the speed at which your mouth can produce a sentence someone else built. It never trains retrieval, building the sentence yourself under time pressure, which is where most of my students fail. You are a very committed echo, as Sarah put it, and nobody asks an echo for its opinion.
I said I was biased about our free page; the same applies here, because I write for Lucida’s blog. Lucida is an AI speaking coach app for iOS and Android: you hold real spoken conversations with an AI coach and get real-time feedback on pronunciation, grammar, vocabulary, fluency, pace and filler words. The coach asks you things. The honest pairing is a short shadowing block for delivery, then a conversation in which nobody hands you the sentence.
A setup for each budget
No rankings; the right stack depends on your language and what you already pay for. Free: Lucida’s shadowing page for English or Spanish, or Apple Podcasts transcripts in any of the six languages, plus Voice Memos on a second device, and YouGlish for single words.
A cheap subscription: shadowing.app’s Standard tier if you are learning English on a computer and want a number, or, at no cost, Speechling’s free plan for Spanish, Italian or Portuguese, where its free coaching applies. Speak Pro belongs here only if its unpublished subscription turns out to be cheap. Check the vendor’s current tiers, not mine.
Already paying for streaming: Language Reactor on Netflix or YouTube, replay-the-line and auto-pause as the loop, dual subtitles as the transcript you hide by switching them off, and a recorder beside the laptop, so this is the stack for shadowing a drama rather than a podcast. Whatever you pick, add conversation practice, or the rhythm you train never meets a sentence you built yourself, and keep one dated recording a month of the same 45-second clip. That file is the only honest progress report a shadowing tool can give you.
Frequently asked questions
Is there a completely free shadowing app?
Yes, with limits. Lucida’s shadowing page is free with no sign-up, in English and Spanish. ShadowingEnglish has free practice pages, Speechling’s free plan includes its curriculum and side-by-side recorder, YouGlish costs nothing, and Language Reactor’s core is free. What paid tiers mostly buy, as of September 2026, is AI scoring and a bigger library, neither of which shadowing needs.
Which shadowing app works for Spanish, French, German, Italian or Portuguese?
Speak Pro lists Spanish, French, German and Italian; Speechling covers all five with a recorder; LingoSon covers all five without one; Language Reactor covers 27 languages and YouGlish more than 25. Most apps with English in the name are English-only, so a Spanish learner searching for a shadowing app is mostly shown the wrong products. Apple Podcasts transcripts cover all five.
Do I need to record myself when shadowing?
Yes, and it is the feature to check before you pay for anything. Play your take beside the original and listen for three things: word endings you dropped, places where the speaker linked two words and you separated them, and pauses that fall where theirs do not. Those are the gains the research supports, and none shows up in a percentage.
Can I shadow with Spotify or Apple Podcasts?
Yes. Apple Podcasts generates time-synced transcripts for shows in all six languages Lucida teaches, and both apps have a speed control; Spotify also skips 15 seconds back. Neither records you, so keep Voice Memos or a second phone within reach. The podcast section above gives the steps, and they are the same in every language.
How many times should I repeat the same clip?
Five or six, then change the text. That figure comes from Shiki and colleagues (2010), who found a plateau after that many repetitions, as summarised by Hamada. Inside those passes you will want one awkward clause eight times over, which is why a loop button matters more than a large library.
Does Lucida have a shadowing mode?
The app does not, and I would not pretend otherwise. Shadowing lives on Lucida’s free web page, with English and Spanish passages, speed control and a recorder. The app does the thing shadowing cannot: a spoken conversation with feedback on sentences you built yourself. Use both, in that order.
Shadow the sentence, then say your own
Lucida is an AI speaking coach app for iOS and Android: you hold real spoken conversations with an AI coach and get real-time feedback on pronunciation, grammar, vocabulary, fluency, pace and filler words. It is free to download, with some features requiring a subscription. Shadow for delivery on the free page, then bring the rhythm to a conversation nobody scripted for you.
