Demo: https://www.loom.com/share/a78e713d46934857a2dc88aed1bb100d?...
We started this company after struggling to find great tools to practice speaking Japanese and French. Having a tutor can be awesome, but there are downsides: they can be expensive (since you pay by the hour), difficult to schedule, and have a high upfront cost (finding a tutor you like often forces you to cycle through a few that you don’t).
We wanted something that would talk with us — realistically, in full conversations — and actually help us improve. So we built it ourselves. The app relies on a custom voice AI pipeline combining STT (speech-to-text), TTS (text-to-speech), LLMs, long term memory, interruptions, turn-taking, etc. Getting speech-to-text to work well for learners was one of the hardest parts — especially with accents, multi-lingual sentences, and noisy environments. We now combine Gemini Flash, Whisper, Scribe, and GPT-4o-transcribe to minimize errors and keep the conversation flowing.
We didn’t want to focus too much on gamification. In our experience, that leads to users performing well in the app, achieving long streaks and so on, without actually getting fluent in the language you're wanting to learn.
With ISSEN you instantly speak and immerse yourself in the language, which, while not easy, is a much more efficient way to learn.
We combine this with a word bank and SRS flashcards for new words learned in the AI voice chats, which allows very rapid improvement in both vocabulary and speaking skills. We also create custom curriculums for each student based on goals, interests, and preferences, and fully customizable settings like speed, turn taking, formality, etc.
App: https://issen.com (works on web, iOS, Android) Pricing: 20 min free trial, $20–29/month (depending on duration and specific geography)
We’d love your feedback — on the tech, the UX, or what you’d wish from a tool like this. Thanks!
Now, I tried the web app and chose to learn Greek as a beginner. And while I had better experience with your app than with ChatGPT or Gemini voice modes, I still got lost 5 minutes in because the AI tutor doesn't seem to have a plan for me, nor does it "see" my struggles. For example, after asking me about a hobby, it gives me a long sentence in Greek about how how it is nice to hike in mountains. Being absolute noob I cannot reply to it, nor even repeat it. And I don't even know what it is expected from me at the moment. A human tutor here would probably repeat a part of the sentence with a translation and ask me to repeat, or would explain something. The AI just sits there waiting for me to make a sound, and when I make it, it goes on on a tangental subject of beach vacations. :)
Again, this is still relatively not bad, and I'm going to give it another try.
My thinking is - I can have unstructured conversations with Advanced Voice Mode or in real life here in Sweden. What I'd really appreciate is a guided learning experience taking me up from intermediate/slightly above intermediate to fluent in the most efficient possible way (as opposed to just having us 'ramble' about random topics of my own choosing).
This is available for all proficiencies. It's just much harder to talk for hours in a new language as a beginner. It's usable but requires more effort.
* curriculum, completely customizable, with grammar, roleplay, topics, speaking speech, transcript, dictionary, corrections, etc
* prompting and AI models all chosen to be a better fit for multilingual, easy to understand, etc.
* the tutor actively tries to teach you, it's not an assistant
* integrated flashcards that go hand in hand with the speaking immersion
I tried following the modern Japanese track on Memrise and was appalled at how bad it is nowadays.
- the name is bad. Issen? I (German, Spanish speaker) don't know how to pronounce it.
- use correct flags. Catalan speakers will be rightfully pissed when you use the Spanish flag for their language.
- in a language learning focused app, it is not acceptable to have a badly translated app. I'm using it in German and while the intro does not have typos, I can tell it's all just AI blubber
- For German specifically, I'd recommend you to use "du" and not "Sie" ("Wie heißen Sie?") across the app. If your tool isn't aimed at 60+ year old BMW drivers, use "du".
Languages and flags don't mix well.
Feedback: The tutor pronounces some obvious things wrong that contradicts the words. Two examples: 気滅の刃 - it pronounced 刃 wrong despite the furigana being correct. It also kept pronouncing は as "ha" even when used as topic particle in more complex sentences. Edit: also observed 使い方 pronounced "saifou" - no idea what's going on there. It was in a mixed english-japanese message.
I think I would pay for this if I wasn't worried about learning mispronunciations or errors.
Oh, more feedback: focus the app on the conversation with the tutor and leave the memorization to Anki - just let us export those words we struggle with to CSV or something so we can import into existing vocab workflows.
Today, it’s not only easier than ever to launch a platform to challenge Duolingo, but its core product—its crowd-sourced human translation service—has been distrupted.
This morning, I found myself thinking about how all those decade-old learning platforms—like Coursera, as reflected in its ever-falling stock price—are being distrupted.
Your product looks awesome and I hope you distrupt all the language learning platforms. Thank you for sharing.
(I had ChatGPT fix my grammatical errors and now this comment doesn't sound like me, sorry.)
And it didn't correct "distrupted" to disrupted?
And mid 2010s view was MOOCs were supposed to disrupt University education!
Add it to the pile.
Why on earth are you going to build out a whole product: doing marketing, security, incorporation, customer support, etc… just so you can finally arrive at the end result of… teaching yourself a language?
The toy app is 100% of the value. You don’t need all that other shit, you just need a really good prompt. It’s exactly how you don’t need to search websites anymore, just ask AI for the answer and get it immediately. You don’t need a full app, just ask AI exactly what you want.
What I usually do is pick a random blurb in the news and paste the entire thing along with the Reuters link at the beginning and inform ChatGPT that we'll be carrying on language practice specifically over that topic of discussion.
I've used this to carry an hour long foreign language practice in Spanish while walking my husky. Just put the phone in my pocket and go. If you're an intermediate/advanced learner, it's a pretty decent solution.
In fact, you can actually instruct ChatGPT that you are going to speak in your native language, but ChatGPT is only allowed to respond in the target language if you just want to focus on practicing listening comprehension.
I'd be interested in hearing how significantly improved Issen is over this.
You do need an app to create a holistic learning experience just for language learning. Customized curriculum, tons of prompting, AI models chosen for transcription accuracy, flashcards/dictionary, etc.
We also support hands free mode, and many other things are customizable like slang, speaking speed, target language usage, etc.
It makes me think the audio goes through a kind of voice-to-text model before the answer, so nuance is lost; or the model wasn't trained to distinguish between correct and incorrect pronunciations.
Does Issen have this issue too? Pronunciation vices are common when you're learning a new language.
The models are improving though, and they are at a very good place for English at the moment. I expect by next year we will switch over to full voice to voice models.
Thank you so much for this. Duolingo is literally unbearable because it's so gamified. I'll try it out later. I've seen a few of these apps, can I seamlessly go between my native language and the language I'm trying to learn? If I am trying to learn Hindi, can I ask a question in English in the middle of a conversation?
These kinds of learning apps are destined to become mediocre over time.
The learning metric is so easy to capture, the learning content so easy to produce, yet no one has an individualized loop to make learning work well.
For example, I'd press "Training" on Duolingo, and would get nowhere. Same lessons all of the time. Bread and water.
Also, for the transcription it would be great to get pure romanji to start with!
The promise and potential of LLM based language learning apps is that you can cross that gap to full immersion in a way that has never been possible before.
Please be more ambitious.
---
AI: Anh mệt is good if bạn are a man speaking about yourself. You can also say, “Em mệt” if you’re a woman.
this isn't correct. If you are of "older brother" age and are male, you say Anh. Em is for if you are "younger person" (does not matter the gender). Women tend to prefer being called "em" (even if they are older), because women prefer to be identified as younger than their true age... But that doesn't mean you can't call younger men em.
A good tutor would know your age relative to theirs and explain this context.
---
It would say english phrases with a vietnamese accent.
---
It also would give me really complex vietnamese phrases that I am not ready for. when I prompt for an explaination or translation, it would get off track from the original thing we were learning.
---
Way more people in Vietnam (and the globe) speak southern Vietnamese, but the tutors seem to be from north Vietnam.
---
The STT also was very forgiving if I pronounced things incorrectly. Or it would confuse english and vietnamese. I would say, "Phai", but it heard "bye"
---
I was ready to pull out my credit card, but I can't trust it to teach me the right information. I pay $160/mo for Vietnamese tutoring ($20 per class). This would be way cheaper and I don't have to schedule my classes.
I think this company will end up pivoting into a B2B context before long. Hopefully they will still stick to the mission, but who knows (and I wouldn't fault them if they don't – survival comes first).
If OpenAI puts resources to language learning, they could build a great product. But 3rd party devs relying on someone's tech hasn't proven to be a good strategy.
The feature request I make for all language course makers: please consider Bengali support in the future! It's wild to me that the 7th most spoken language in the world, with a deep culture around literature and poetry [1], gets zero attention from language course makers. I can buy an Assimil course on Breton, spoken by 200k people, and not Bangla, spoken by 284 million.
[1] https://en.wikipedia.org/wiki/Charyapada
Suggestion: you may be able to integrate SRS into the conversation. —- you could encourage the model to use certain words, and more importantly you can track the student’s active use of words that are on the review list, basically acting as if it were an SRS step. — this could totally eliminate the need for flashcards.
Don’t dim screen on iPhone during conversation.
The tutor should terminate the lesson when its goals are achieved and do a warm handoff.
Overall it’s quite good.
I've been learning Arabic, and I noticed that the app uses Arabic script right from the start. This can be quite challenging for beginners who haven't learned how to read it yet. May I suggest adding an Englishized (romanized) version of the Arabic text to help ease the learning curve?
It also seems to not listen to me when I asked to give me shorter sentences. It seems to not care that I'm struggling despite my pleading.
I later switched to Spanish, which was a better experience. This one seems to listen to me better. I can ask the tutor to repeat what they said in English and give me shorter sentences, and thankfully, it does.
Interacting with the tutors does feel I have to drive the conversation which is taxing. Compared to a human tutor, where I feel assured that I can be guided properly.
Still an interesting app. Would love to try Spanish some more, in the future.
Reading in a non-familiar script becomes much easier the more you do it, and the longer you put off learning it, the more opportunities you miss for using it.
I think you should only be using the latinized scripts in the absolute beginning where you are learning the most basic words and phrases like: “hello”, “yes”, and “no”, and “what is your name?”. This should only be for your first couple of weeks. After that you should have learned to read new words in the new script (albeit slowly). Learning the script makes everything much easier afterwards.
However, gamification can only do so much and I'm afraid language learning is a lot like learning to code: many people want to want it but few actually want it. In that case, presenting as a "want it" when you are a "want to want it" is social proof and largely unrelated to whether you are actually learning (as long as the pretense is kept up) — hence the success of Duolingo despite the relatively poor real-world outcomes. In Duolingo's case the streaks are even explicitly considered to be social proof.
For pitch accent, shadowing is a great way to improve. You can pause and repeat the tutors messages for example, or read out the word when doing flashcard reviews (copying the flashcard audio).
These "we cover every single language" tools get it like 75% right at best.
It's a browser extension that finds English sentences in webpages, and translates the ones at your difficulty level into the language you're learning.
Or something that tries to teach 60 languages but does so poorly ?
Even some basic app that can pre-load the prompt doesn't seem to exist?
I think showing the raw reasoning text is not quite the right UI; maybe highlighting the specific text in red and showing a suggested correction would work better?
It's also a little awkward that the conversation is live; I don't really have any breathing room to read the reasoning traces on what mistakes I made / could have done better. I hung up the first time I tried to figure out how to pause.
I’m going to assume this works better on the App.
- Don't go over 10k tokens in the prompts as the intelligence and memory degrades
- Summarize sessions and save the summaries, potentially summarize the summaries as well
- use VAPI or realtime api if you want to build fast. Building the full pipeline takes a while
- try out different models and see how personality varies. Our favorite is gpt4.1 with temperature 1.
- goal system. The promot should always contain the current goal, and the next goal. Evaluate goals with another LLM, and dynamically change the prompt
Part of me loves this—no judgement, endless convenience, cheap. But another part mourns, sensing it strips away the grit, the stumbles, the soul of language learning. The kind that only comes from fumbling through conversations with another human.
When I was learning Spanish, I used italki extensively and found having a live Columbian tutor invaluable and very affordable for most Westerners. It would genuinely make me sad if those excellent tutors start losing work to this kind of AI.
I tried the Japanese track. I'm a total beginner and the first lesson wasn't helpful at all. The AI asked about maybe mixing up Japanese<>English, but it didn't actually follow through. It either spoke fully in Japanese or fully in English. Maybe this is a standard practice for language lessons? I remember going to the first day of French class in a community college, and the teacher only spoke French, which was extremely overwhelming. Perhaps it's the standard way of teaching? Even if it is, I'm not sure if it works when compressed down to the shorter times I see myself opening the app.
You can delete your account at any time to fully wipe all your data, but there is no way to delete sessions ATM.
For me a key feature will be a family plan; Duolingo is great in that regard.
That person will have a hellish time marketing it because projects like this will have so thoroughly primed us to assume its slop.
"Error An unknown error has occurred."
has this been tested on Safari?
My first problem was setting my native language truthfully to Icelandic which seemed to confuse both me and the AI tutor. We spoke together in Japanese but asking how to say a word in Japanese but giving the Icelandic word didn’t quite work, giving the word in English worked much better.
Now as a beginner I don’t think this service is right for me. It is very hard to have a conversation—even a basic one—at my level and I didn’t actually learn that much as I wasn’t able to say anything. I did however learn that I need to practice creating sentences on my own, and I need to practice speaking, but honestly I would much rather do that via structured exercises from a textbook then from an AI tutor (or a human tutor for that matter). I have been skipping those exercises in the textbook that I use, so I guess having that 20 min conversation did indeed help me realize what I need to focus on. So I guess thanks for that.
A more useful feedback from a beginner’s perspective. Taking your time between sentences is something you can do with an AI tutor which you can‘t do with a human tutor, so I recommend you add stuff like dictionaries and grammar keys which beginners can look up before starting the next sentence.
I would also like to see some basic note-taking, or even message drafting, such that you can type in a draft before you start speaking your next sentence. I don’t think intermediate speakers would need these as they can just ask the AI tutor during the conversation, but for beginners it is nice to have some written materials as you practice.
How widely have you tested your supported languages on native-speakers and learners?
We've done a lot of testing on Spanish, English, Italian, Japanese, and French, but much less on the others and none at all for some of the niche ones.
The language support is based on the intersection of the languages that have low word errors rates in the transcribers, as well as officially supported by LLM/TTS (like gpt4.1, eleven labs etc).
We've seen the models' quality improve consistently over the last 6 months, in all languages we tested, and now the error rates are getting really low.
Certainly if your product were to mis-teach me important details, and I were to then find out that you had spent less time testing than I had spent learning, I would be quite angry.
The vocabulary tooling looks neat and well thought out.
Also, we prefer the eleven labs voices, but there is definitely varying quality. I'm guessing later this year or next, the voice to voice models will become good enough, and we will switch over.
The teacher kept switching into an American accent when I was trying to learn French and the responses were getting very slow bitty.
Hopefully this is just an initial load of issues because the concept is great.
https://x.com/JarrettYe
The faq wont expand on tap for me on android firefox. Dm me if you need more info.
Looks like a great app and I can't wait to try it for Japanese!
Can the cards be exported to anki?
The thing that put me off was the speech recognition. I am not in a loud environment and I wasn't even talking and it was picking up responses and responding to it before I even opened my mouth. It blazed through the 'preferences' set up itself making up responses. Then when I did get to talk it just simply got my answers wrong. It would often interject too at random during my sentences.
Whisper can transcribe in <100ms. We then wait for the turn detection model, LLM, and tts to trigger a streamed response back to eh client.
I’m currently learning French as a beginner and I’ve learned other languages in the past. I’ve trued Duolingo as well as italki and frantasic as well as just ChatGPT. I am very familiar with Anki and I think it’s critical to make your own flashcards by choosing images and sounds. I don’t want auto cards.
My experience with Issen:
* it’s frustrating when the conversation partner doesn’t remember what it just said - it means I can’t get a chance to ask que c’est que ça veut dire.
* it’s frustrating (just like with ChatGPT) that the conversation partner tends to interrupt and jump in while I’m thinking. I think many learners speak slowly and spend extra time thinking. ChatGPT allows you to hold the glowing circle and it won’t interrupt while you do.
I’d love to see the chat bubbles have more in depth features like:
* much clearer indicator of hover or click words for translation, and more features like example sentences or click to pronounce
* an option to ask for an explanation of some or all the text
* for my own text I’d love to see feedback with more UI native elements about how accurately I pronounced each word and any grammatical mistakes I made. The text summary is a great start
I found myself ignoring the features of the chat bubbles and only in writing this feedback did I notice them! They could maybe use more contrast and clear UI emphasis. Duolingo does a good job of making their UI very clear with this kind of feedback.
I think it’s important to build features that augment the app to work around LLM limitations. My guess is a lot of the settings change the prompt and that’s great but I think it leaves too much room for hallucinations to nosedive the experience.
I’d also love to see some way to have a hold to talk or something similar.
I’m very conscious at this point about the cost of these lessons and I have a hard time finding the price. Frantastic is absurdly expensive and it made me switch to italki where human conversation is literally cheaper. Without differentiating more from ChatGPT I would have a hard time justifying an additional subscription to my wife!
Edit: I found the pricing and it’s a tough sell! ChatGPT is cheaper.
I think you can both differentiate further from ChatGPT and keep cost down. I’d recommend to try to get more value out of each API call, so learners are more aligned with the cost per interaction- like make it so I’m enticed to spend a little longer reviewing the chat bubbles. My suggestions are mostly about how I want more engagement with each utterance anyway. Right now it’s very tempting to just keep making more and more utterances and IMHO that drives up costs while being frustrating for me.
I’d be happy to discuss! I wish you success.
In terms of the exact resources for pronunciation - The Fluent Forever guy has a good anki deck for $12 (I bought it and I'd recommend it - just have patience and know he tends to over explain IMHO but the cards are linked in there and they're great) https://blog.fluent-forever.com/chapter3/ and I'd recommend finding your own favorite YouTube videos to explain how to pronounce the French R and nasal sounds. I would try watching some YouTube in French just to wet your beak. Know that it's frustrating to not yet have good comprehension but keep at pronunciation/comprehension and you'll get there.
I recommend making Anki cards for like the top 100 and then the top 500 words, and include images and sounds (Anki strengths).
I'd suggest to have a goal of understanding some rewarding things like children's T.V. (Bob l'éponge) or language learning YouTube (Easy French) - really fun. Then after you master some early words and feel like you have a "French ear" jump in and do some "early reader" kinds of book (https://www.barnesandnoble.com/w/short-stories-in-french-for...) because that will be really rewarding and reenforcing.
I also recommend jumping in to italki probably earlier than you feel comfortable (or this app, as it continues to improve!) and doing some community conversations in just an unstructured way. Just be ready to try a couple people and find someone you like. If you can travel to France I think that is probably best, too! You'll be very happy that you've got a good "R" at this point.
I think at that point you're ready to look at the A1/A2/B1/B2 test content and learn it on your own pretty easily or work with a structured tutor. It should be chill and not too challenging at that point.
It asks me for my level; I'm half way through this audiobook (https://www.amazon.co.uk/Next-Steps-Spanish-Paul-Noble/dp/B0...), and have listened to the book before it a number of times, so I'd say I'm between beginner and intermediate. I think you could do better than a "what level are you, pick from 3 options" and throw you straight into a chat - ask some basic Spanish questions, and then try and figure out where the user is from there.
Next I chose Blanca from Barcelona, and she said an awful lot of words and I understood very little of them, so I think I'm not ready. Half the grammar lessons have both a Spanish and English explanation, and half don't.
I'll keep the feedback coming, but I'm on a train now and the questionable internet is not good enough for an actual conversation.
(not at all relevant but I work for Devyce, from the YC S22 batch!)
You can't make up a couple of conversation topics and expect the LLMs to do the rest by just switching languages. People approach the same topics completely different in different languages. The app looks like someone picked a couple of topics and the rest is "just" ChatGPT advanced voice mode.
And the worst thing is that the LLMs in TTS do not sound native and cannot teach you pronounciation and learning to listen and understand (which is the whole point in having spoken conversation).
And the other way around, the STT will not notice pronounciation mistakes made by the student - so the app cannot tell you: oh, its pronounced like this.
If AI succeeds, will we even need language learning? Language learning is notoriously difficult and time-consuming. We're rapidly approaching a future like Star Trek of universal translators.
If so, what is the realistic future for AI language learning products?
Plus, you can't do auto translation for languages like Japanese where the grammar is reversed. Auto translation has fundamental limitations.