Understand an argument from the real recording, not from memory

An AI argument analyzer for couples: who said what, and how you both communicated.

This isn't telling a chatbot your side of the story: FeelPair works from what was actually said, with each person's timing. It doesn't pick a winner; it helps you see what each of you was asking for and how to pick it up again. Only you see it, and your partner doesn't need an account.

Try it now, no account needed

Audio file

MP3, M4A or WAV. Up to 5 minutes and 64 MB.

Without an account you can try one recording of up to 5 minutes. With a free account, up to 20.

Language you are speaking Date of the conversation

Taken from the file name. If the file carries its own date, we use that. You can change it now or later.

I am one of the two people speaking and we are both adults. I am not uploading other people's conversations. The other person knew it was being recorded. I am not uploading a secret recording. I agree that the audio is sent to OpenAI to be transcribed, and the reviewed text for the reading. It is not used to train models.

Only upload conversations where the other person knew they were being recorded. Without an account, the recording is deleted after 24 hours.

How the argument analyzer works

1
Upload the audio of the conversation (MP3, M4A or WAV).
2
It's turned into text and the two voices are separated, with the moment of every line.
3
Review it, fix anything that's off and say which voice is yours.
4
Get a reading of how you communicated and, with a free account, keep talking it through.

What you get

Who said what, and when

The whole conversation as text, split by voice, with the minute of every line. Listen to any part and correct it.

What can be counted

How long each of you talked, how many times, how many questions you asked and who started talking over the other. Numbers, not judgments.

A reading with exact quotes

What each of you seemed to be asking for, where it got stuck, the attempts to reconnect, and another way to say your own lines. Every observation quotes what was said.

Keep talking it through

With a free account, a chat about that conversation: how to bring it up again, what to ask, how to say what was left unsaid.

Why not just tell ChatGPT about the fight?

When you describe an argument to a chatbot, the AI only has your memory of it, and general chatbots tend to agree with whoever is asking. Here the starting point is the recording: what was said, in what order and how much each person talked.

And the reading is built to be fair to the person who isn't there: it reads your partner's words as generously as yours, never changes what they said and never builds a case against anyone.

What it doesn't do

It doesn't say who was right.
It isn't evidence against anyone, and it isn't a way to monitor your partner.
It doesn't replace a professional. If there is fear, threats or violence, it gives you places to get help instead of a reading.
It doesn't guess tone of voice: it works with the words.

Privacy

Only you see the recording, the transcript and the reading. Your partner sees nothing, whether or not they have an account.
Everything is stored encrypted. Without an account it's deleted after 24 hours; with one, the audio after 30 days and the text after 180. You can delete it sooner whenever you like.
It isn't used to train models.
Only for conversations where the other person knew they were being recorded.

Frequently asked questions

Will it tell me who's right?

No. No recording is enough for that, and FeelPair doesn't give verdicts. It shows what each of you was asking for, where the conversation got stuck and what you could do differently next time.

Did my partner need to know I was recording?

Yes. Before uploading you confirm that the other person knew they were being recorded. In many places recording a conversation without everyone's consent is illegal, and FeelPair isn't for secret recordings.

How is this different from telling ChatGPT about the fight?

A chatbot only hears your retelling. FeelPair starts from the recording itself: who said what, when and for how long, and every observation quotes the exact words. It's also built to be as fair to your partner as to you.

What files work, and how long can they be?

MP3, M4A or WAV files of a conversation between two people, such as phone voice memos. Up to 5 minutes without an account, up to 20 with a free one.

Who can see my recording?

Only you. Your partner doesn't need to use FeelPair and sees nothing you upload. It's stored encrypted, deleted after 24 hours without an account, and never used to train models.

Can I fix wrong words or speakers?

Yes. You can listen to every line, correct the words or who said them and leave parts out. What was recognized at first stays saved, and the words you correct help transcribe you better next time.

What if we keep having the same argument?

The reading shows what happened in that conversation: who asked, who answered, where it got stuck. With a free account you can keep talking it through and prepare a different way to start the next one.

How much does it cost?

Trying it is free and needs no account. With a free account you keep the recording, see the full reading and keep talking it through.

Do you have the audio of that argument?

Upload it and in a few minutes you'll see who said what and how you communicated. Up to 5 minutes without an account.

Try it free