Decode Their Texts

A free tool that analyzes a text conversation and surfaces the emotional patterns inside it. You paste the messages (in the format `Name: message` per line), and the decoder spots signals like emoji frequency per person, message length imbalance, one-word replies, trailing ellipses, and known slang combos (🥺👉👈, 😎🔥, 💀💀💀). It outputs labeled "vibe readings" with a confidence rating, plus per-person stats. No AI, no API call — pure pattern matching that runs in your browser.

The decoder runs a set of hand-curated rules over your conversation. Examples: "if person A sends 1.8× more messages than person B, that's a lopsided-engagement signal," or "if 🥺 is someone's top emoji and they used it 3+ times, that's an anxious / seeking-reassurance pattern." It also detects 50+ specific emoji combos from our reverse decoder (🥺👉👈 = shy asking, 👁️👄👁️ = awkward witnessing). Every reading carries a confidence label (High / Medium / Low) so you know how seriously to take it.

It's pattern-matching, not mind-reading — and the disclaimer at the bottom of every result says exactly that. What it can do honestly: spot statistical mismatches you might miss when reading one message at a time. If you send 7 emojis per message and they send 0.3, that mismatch is real and meaningful. If they reply with one word 60% of the time, that's a signal. The decoder surfaces these patterns objectively. What it cannot do: tell you definitively whether someone "likes you back". Use it as a lens, not a verdict.

Format each message on its own line as `SenderName: message text`. Example: `Me: hey, how was the party 🥺` then on the next line `Alex: it was ok lol`. Multi-line messages are merged automatically — only lines that start with a Name + colon are treated as new messages. Timestamps in iMessage / WhatsApp format ([12:34 PM] Sarah: ...) are also parsed. iMessage screenshots don't paste as text, so you have to type or copy the messages manually.

Per person: total messages, average message length, emojis per message, top 5 most-used emojis, all-caps word count (emphasis / shouting), question marks (curiosity / engagement), trailing ellipses (hesitation), one-word replies (disengagement). Whole conversation: emoji density, message-count imbalance between speakers, emoji-energy mismatch (one warm, one cold), and presence of known slang combos. From these we generate up to 8 vibe readings.

Nowhere. Everything runs in your browser tab — no server call, no analytics on the conversation text, no logging. Once you close the tab, the analysis is gone. This is a pure client-side tool: open the page, paste, get your read, leave. We never see your messages.

Reading messages one-by-one, you naturally project your hopes onto them. The decoder gives you objective counts: "they sent 3 one-word replies in 5 messages" is a number, not a feeling. It's especially useful when you're deep in a situationship and can't tell if you're reading too much into things. The numbers don't lie about the patterns; they just don't tell you what to do with them.

Yes. After the decoder runs, the "Copy analysis to share" button puts a clean text summary on your clipboard — perfect for sending to a friend for a second opinion, or pasting into a group chat for a "wait what does this mean" discussion. The shared text never includes your original messages, only the readings and stats.

Used once: cute, soft, often paired with asking for a small favor or reacting to something sweet. Used multiple times in one conversation: usually a "seeking reassurance" or anxious pattern — the sender wants validation that things are okay. The decoder flags this when 🥺 appears 3+ times as a person's top emoji. It pairs especially with 👉👈 (shy asking) and 💕 (overwhelmed-by-love).

That's the ideal case — the decoder runs additional comparison checks: who sends more messages, whose emojis are warmer, who asks more questions. With three or more speakers, you still get per-person stats and combo detection, but the comparison readings are skipped because they get noisy with too many people.